You are on page 1of 12

Cancer Letters 325 (2012) 42–53

Contents lists available at SciVerse ScienceDirect

Cancer Letters
journal homepage: www.elsevier.com/locate/canlet

Activation of canonical WNT/b-catenin signaling enhances in vitro motility of


glioblastoma cells by activation of ZEB1 and other activators of
epithelial-to-mesenchymal transition
Ulf D. Kahlert a,h, Donata Maciaczyk a, Soroush Doostkam b, Brent A. Orr c, Brian Simons c, Tomasz Bogiel a,
Thomas Reithmeier a, Marco Prinz b,f, Jörg Schubert d,1, Gabriele Niedermann g, Thomas Brabletz d,
Charles G. Eberhart c, Guido Nikkhah a, Jaroslaw Maciaczyk e,⇑
a
Division of Stereotactic Neurosurgery, Department of General Neurosurgery, University Medical Center Freiburg, Breisacher Straße 64, 79106 Freiburg, Germany
b
Department of Neuropathology, University Medical Center Freiburg, Breisacher Straße 64, 79106 Freiburg, Germany
c
Department of Pathology, Johns Hopkins Medical Institutions, Smith Building, 400 North Broadway, Baltimore, 21287 MD, USA
d
Department of General- and Visceral Surgery, University Medical Center Freiburg, Hugstetter Straße 55, 79106 Freiburg, Germany
e
Department of General Neurosurgery, University Medical Center Freiburg, Breisacher Straße 64, 79106 Freiburg, Germany
f
BIOSS Center for Biological Signaling Studies, University of Freiburg, 79106 Freiburg, Germany
g
Department of Radiation Oncology, University Medical Center Freiburg, 79106 Freiburg, Germany
h
Faculty of Biology, Albert-Ludwigs University of Freiburg, Schaenzlestr 1, 79194 Freiburg, Germany

a r t i c l e i n f o a b s t r a c t

Article history: Here we show that activation of the canonical WNT/b-catenin pathway increases the expression of stem
Received 15 March 2012 cell genes and promotes the migratory and invasive capacity of glioblastoma. Modulation of WNT signal-
Received in revised form 15 May 2012 ing alters the expression of epithelial-to-mesenchymal transition activators, suggesting a role of this pro-
Accepted 22 May 2012
cess in the regulation of glioma motility.
Using immunohistochemistry in patient-derived glioblastoma samples we showed higher numbers of
cells with intranuclear signal for b-catenin in the infiltrating edge of tumor compared to central tumor
Keywords:
parenchyma. These findings suggest that canonical WNT/b-catenin pathway is a critical regulator of
Glioblastoma
b-Catenin
GBM invasion and may represent a potential therapeutic target.
ZEB1 Ó 2012 Elsevier Ireland Ltd. All rights reserved.
Epithelial-to-mesenchymal transition (EMT)
Invasion

1. Introduction capacity of BTSC appears to be regulated by developmental path-


ways including Sonic Hedgehog [5], Notch [16] and possibly Hippo
Glioblastoma multiforme (GBM), the most malignant primary [40] signaling. Interestingly, while WNT/b-catenin signaling has
brain tumor is characterized by dismal prognosis with median sur- been implicated in the proliferation of the neural stem cells and
vival of approximately 16–19 months despite multimodal treat- the regulation of stem-like cell population in a number of solid tu-
ment including surgical resection followed by combined radio- mors (for review see [10,39]), only recently has it been discussed in
chemotherapy [53,54]. In spite of enormous efforts towards under- the context of GBM [3,24,43,62,64].
standing the molecular basis of the disease [2] and the develop- Mutations of pathway members, including adenomatous polyp-
ment of novel therapeutic strategies only limited advances have osis coli (APC) or b-catenin itself result in aberrant activation of the
been achieved. However, the identification of brain tumor stem- target genes, including those encoding for activators of epithelial-
like cells (BTSCs [19,51]), in both adult and pediatric human malig- to-mesenchymal transition (EMT). EMT has been shown to pro-
nant gliomas, have provided new insights in the biology of malig- mote mesenchymal differentiation and migration in a number of
nant glial tumors rendering these cells a potentially viable human tumors including, i.e. carcinomas of the colon [65], prostate
therapeutic target. BTSC share many characteristics of normal neu- [36] and breast [13].
ral stem cells. Similar to normal neural stem cells the proliferative Integrated genomic analysis enabled the molecular classifica-
tion of GBM into neural, proneural, classical and mesenchymal
⇑ Corresponding author. Tel.: +49 761 270 63570; fax: +49 761 270 63580.
subtypes [56]. Compared to the more favorable proneural subtype,
the mesenchymal signature is associated with poor response to
E-mail address: jaroslaw.maciaczyk@uniklinik-freiburg.de (J. Maciaczyk).
1
Present address: Novartis Pharma GmbH, Roonstr. 25, 90429 Nürnberg, Germany. therapy and worse prognosis [41]. Interestingly, evaluation of

0304-3835/$ - see front matter Ó 2012 Elsevier Ireland Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.canlet.2012.05.024
U.D. Kahlert et al. / Cancer Letters 325 (2012) 42–53 43

morphometric details (i.e. nuclear staining) from the TCGA data- (S33Y) and therefore missing the phosphorylation site subsequently leading to
aberrant turn-over of the protein with arrested ubiquitination, cytoplasmatic accu-
base revealed that high levels of WNT/b-catenin activity is associ-
mulation and high nuclear transfer. We cloned the mutant b-catenin under the con-
ated with shortest overall patient survival [11]. This may be due to trol of an EF1alpha promoter into the cFUGW lentiviral backbone, containing a GFP-
increased cell motility leading to early tumor recurrence, processes sequence as a selection marker, driven by human ubiquitin promoter. This dual pro-
that may be dependent on EMT [33,60]. moter system appears to be constitutively on in all tissues and rarely undergoes
In this study, we present evidence of canonical WNT/b-catenin silencing events [30].
For overexpression of Dickkopf1 (DKK1) we cloned the sequence from pCS
pathway activation predominantly within the invasive tumor front
hDKK1 (Addgene #15494) coupled with a CMVpromoter into the lentiviral back-
in GBM samples. Activation of the pathway in GBM-derived cul- bone of pCDH-CMV-MCS-EF1copGFP (System Biosciences). Here the selection mar-
tures induced expression of multiple EMT activators including ker is a GFP-sequence from Pontellina plumata (copGFP).
ZEB1 [57], Twist [33] and Snail [60] and augmented in vitro cell In case of b-catenin-knock down experiments we applied short hairpin (sh) RNA
transfection method. To block the intrinsic b-catenin signaling we applied the
migration/invasion. Conversely, inhibition of the pathway using
pLKO.1.sh.beta-catenin plasmid [17] (Addgene plasmid #1248) and its control
RNA-interference, led to down regulation of EMT-inducing tran- pLKO.1.sh.scramble plasmid [46] (Addgene plasmid #1864) both obtained from
scription factors and decrease of in vitro cell motility. Addgene Cambridge, MA.
Our findings support a novel role of the canonical WNT pathway HEK 293T cells were cultured to 80% confluence in 10 cm petri dishes. In anti-
in regulating GBM invasion by inducing a mesenchymal develop- biotics free environment, the cells were transiently transfected with 16 lg plasmid
DNA using Lipofectamine 2000 (Invitrogen). The transduction of >90% of the cells
mental program and suggest that WNT/b-catenin pathway may
was checked under fluorescence microscope. The virus was collected 48, 72 and
represent a potential therapeutic target. 96 h after transfection and concentrated via ultracentrifugation using ultra-speed
centrifuge (Beckman Coulter) at 22.500 rpm/min. Each virus production from a
2. Materials and methods 10 cm2 dish was concentrated in 100 ll PBS, aliquoted and stored at 80 °C. The
virus titer was determined by quantitative PCR according the protocol described
2.1. Clinical data and tissue collection by Sastry et al. [48].

All procedures described were approved by the local ethical committee of the
2.5. Lentiviral infection of target cells and positive selection for infected cells
University of Freiburg and were in accordance with the German law. For immuno-
histochemical analysis, tumor tissue from 30 patients diagnosed with glioblastoma
To infect the GBM cultures (suspension as well as adherent) we plated
in accordance with World Health Organization established guidelines [31] was used
3  105 cells in a volume of 1 ml culture media in one well of a six well dish. The
for this study.
cells were infected with a multiplicity of infection of 10. After an initial infection
Primary tumor-derived cell cultures (KK, FF, AH and GB) were established
time span of 5–6 h at 37 °C 2 ml of growth media was added. For some cell lines
according to procedure developed for human fetal neural stem cells adopted from
we used two consecutive rounds of infections for S33Y mutation. To obtain a pure
Maciaczyk et al. [32].
population of infected cells with the S33Y-, cFUGW-plasmids as well as for DKK1
Cell lines NCH421k, NCH644 were kindly provided by C. Herold-Mende, Depart-
overexpression and its control we used FACS sorting (100 lm nozzle, MoFlow Sys-
ment of Neurosurgery, Medical Center Heidelberg, Germany.
tem, Dako Cytomation) against high FITC-intensity.
Cell line HSR-GBM1 was gratefully provided by A. Vescovi, Stemgen S.p.A., Mi-
In case of sh-RNA transfections, the Puromycin resistance gene enabled an anti-
lan, Italy, whereas cell line 276 was generously provided by A. Quinones-Hinojosa,
biotic selection of the transfected cell population starting 72 h post infection.
Department Neurosurgery, Johns Hopkins Medical Institutions, Baltimore, USA.
HEK293T cells were provided by M. Bredel, Department General Neurosurgery,
Medical Center Freiburg, Germany. 2.6. Immunocytochemistry
U87 was graciously provided by A. Weyerbrock, Department General Neurosur-
gery, Medical Center Freiburg, Germany. For immuncytological studies, the cells were plated on pre-coated glass cover-
30 GBM samples for immunohistochemical evaluation were obtained from slips in a 24 well plate. Cells were plated as drop culture to enhance their attach-
Department of General Neurosurgery, Medical Center Freiburg (head Prof. J. ment for 3 h. After incubation the cells were fixed in 4% paraformaldehyde,
Zentner). blocked with 4% goat serum and stained with the corresponding primary antibod-
The adult cerebral cortex from a non-tumor patient was gratefully provided by ies. The primary antibody incubation was performed overnight at 4 °C. Afterwards,
C. Haas, Experimental Epilepsy, Department of General Neurosurgery, Medical Cen- following several washes, secondary Alexa 488 (Invitrogen, anti-mouse, A11001
ter Freiburg, Germany. 1:250) was applied for 1 h at room temperature. The nuclei were counterstained
with Dapi (Sigma–Aldrich). (For detailed information on antibodies applied see
2.2. Analysis of expression of WNT target genes Supplementary Fig. S1).

Subgroup specific mRNA expression data (z-normalized) from the TCGA [56]
was downloaded from the cbio genomic portal on February 2012 (http://www.cbio- 2.7. Immunohistochemistry
portal.org). Box plots were generated applying Interactive Box plot tool provided by
the shodor portal (http://www.shodor.org). Specimens of surgical tumor resections (n = 30) from patients with GBM, of
which two exhibited additional sarcomatous components, were immunohisto-
chemically assessed for presence of intranuclear b-catenin (#610153, Becton Dick-
2.3. Cell culture
inson Biosciences, 1:200). In brief, the formalin fixed and paraffin embedded
samples were sectioned to 3 lm thickness. Following deparaffinization, the tissue
Tumors spheres (KK, HSR-GBM1, NCH421k, GB, FF, NCH644 and 276) were cul-
sections were subjected to water steam for 40 min at pH6 for antigen retrieval
tured as described previously [32] in Dulbecco’s modified Eagle’s medium (DMEM)/
and stained with anti-b-catenin primary antibody at 4 °C overnight. To distinguish
F12 (3:1) medium containing 1  B27 (Gibco) supplements. The media was supple-
tumor cells from endothelium, which often shows abundant intranuclear b-catenin
mented with 20 ng/ml recombinant human basic fibroblast growth factor (bFGF,
signal, specimens were stained with anti-CD31 (#JC70a, Dako,) at room tempera-
R&D Systems), 20 ng/ml recombinant human epidermal growth factor (EGF, R&D),
ture. Subsequently, the slices were incubated with secondary antibody (#E464, bio-
penicillin/streptomycin (Invitrogen), L-glutamine (Invitrogen), and 5 lg/ml heparin
tinylated polyclonal rabbit anti-mouse, Dako, 1:200) for 60 min at room
(Sigma–Aldrich).
temperature. b-catenin and CD31 were detected, following incubation with horse
Twice per week the spheres were passaged via mechanic dissociation (low vol-
radish peroxidase (HRP)-conjugated-Avidin (1:1000, 45 min, room temperature),
ume tituration up to 150 times with polished 200 ll pipettes) combined with enzy-
using DAB solution (10 min, room temperature, Merck). Counterstaining was per-
matic digestion with Trypsin Like Enzyme (TrypLE, Gibco) at 37 °C.
formed with Haematoxylin for 30 s.
The HEK 293T cells as well as U87 were grown in DMEM (Gibco) supplemented
Each specimen was evaluated for nuclear b-catenin expression, indicating WNT
with 10% fetal calf serum (FCS) and 1 antibiotics and antimycotics (Gibco).
pathway activation. Sections were divided in central tumor parenchyma (vital tu-
mor center), infiltration/margin zone (progression of tumor core to non-neoplastic
2.4. Generation of lentiviral constructs tissue, with significant decreased cell density compared to tumor core) and adjacent
non-neoplastic brain tissue. In each tumor the percentage of b-catenin positive nu-
For stable transfection we used the 3rd generation of lentiviral packaging sys- clei was assessed. Afterwards, a semi quantitative scoring system was applied
tem (VSV-G; RRE, REV, original distributed by Trono et al. [15]). To overactivate which distinguished 0 – ‘‘no signal’’ indicating a lack of nuclear b-catenin staining,
the canonical WNT/b-catenin pathway we used a mutant form of b-catenin, bearing 1 – ‘‘mild activity’’ corresponding to less than 1% stained nuclei, 2 – ‘‘moderate
a Cytosine to Adenine exchange at position 33 (originally generated by Morin et al. activity’’ for staining present in 1–5% of nuclei and 3 – ‘‘substantial activity’’ corre-
[34]). This transversion gives an altered amino acid sequence for Serine to Tyrosine sponding to more than 5% of stained nuclei for each of three tested regions per sam-
44 U.D. Kahlert et al. / Cancer Letters 325 (2012) 42–53

ple. Colon carcinoma samples were used as a positive control, whereas non-cancer- 2.13. Statistical analysis
ous adult brain served as negative control. (For complete table with pathological
scoring see Supplementary Fig. S2). A statistical analysis was carried out with Statistica v5.1 (StatSoft Inc., USA). A
one way ANOVA test was used to compare the different groups. In case of statisti-
cally significant differences, the Newman–Keuls post hoc analysis was used to iden-
2.8. RNA extraction and reverse transcription tify the nature of the differences.
For q-PCR results, means based on the CP value–raw data for each gene and con-
Total RNA was isolated using the RNeasy Mini kit (Qiagen) according to the dition were compared with Student t-Test using Statview 5.0 (SAS Institute Inc.,
manufacturer’s instructions. Contamination with genomic DNA was minimized by USA).
applying DNA digestion step with DNAse easy set (Qiagen). Total RNA extracts were If not mentioned otherwise, the level of significance was set at p < 0.05 for all
photometrical quantified in a UV cuvette in a Biophotometer (Eppendorf). Reverse tests. Results are expressed as means ± standard derivation.
transcription (RT) was performed according to manufacturer’s protocol using the
SuperScriptÒ III Reverse Transcriptase kit (Invitrogen). Translation of messenger
RNA only was achieved by applying oligo-dT primers. Copy-DNA was diluted to a 3. Results
10 ng/ll working solution. RNA stocks were stored at 80 °C and cDNA at 20 °C.

3.1. Multiple WNT/b-catenin targets are overexpressed in the


2.9. Real-time quantitative RT-PCR mesenchymal subgroup of GMBs

Abundance of transcripts was determined by real-time quantitative PCR on a


The Cancer Genome Atlas (TCGA) recently completed a compre-
MyiQ Real-Time PCR Detection System (Bio-Rad Laboratories) via SYBR Green incor-
poration (Applied Biosystems) for each gene (performed in triplets). Primers were hensive genomic analysis of GBM which included the subdivision
used at a concentration of 10 pmol/primer. PCR conditions were as follows: 35 cy- of tumors into molecular subtypes including the proneural, neural,
cles, 15 min at 95 °C, followed by 45 cycles of 15 s at 95 °C and 1 min at 60 °C each. classical, and mesenchymal subgroup ([38], [56]). To determine if
Melting curves of the amplified products were generated to control for specificity of genes related to the WNT/b-catenin pathway were differentially
the amplification reaction. All relative quantifications were normalized to endoge-
expressed in specific subgroups of GBMs we interrogated the TCGA
nous housekeeping control (b-Actin). In case of down regulated expression a value
of 1 was set as reference level. dataset using the caBio portal (www.cbioportal.org). Besides CD44,
Amplification values were automatically determined by supplied software of a cell surface glycoprotein implicated in Apc/Tcf- mediated tumor
the MyiQ5 Real-Time PCR Detection System (Bio-Rad). (For primer sequences see formation [58], Runx2 a transcription factor responsible for mis-
Supplementary Fig. S3).
regulated differentiation in WNT-mutant chondrocytes and osteo-
blasts [14], the WNT-ligand receptor Frizzled-1 [29], as well as
2.10. Western blot others were investigated. Surprisingly, we discovered significant
higher expression of Dickkopf-1 (DKK1), one of main regulators
Protein analysis was performed with standard semi-dry Western blot method
of the canonical WNT/b-catenin pathway, exclusively in the
under established protocol settings. Samples were analyzed with a loading of
40 lg protein/lane. The primary anti-body staining was done via overnight incuba-
aggressive mesenchymal subgroup of GBM (Box plots for TCGA
tion at 4 °C after blocking with 5% milk powder solution or 2% SlimFastÓ solution. A dataset expression values see Fig. 1). Following this analysis we
list of primary antibodies applied can be found in Supplementary Fig. S1. Secondary investigated the role of canonical WNT/b-catenin network in an
HRP-conjugated antibodies (anti-mouse: #474-1802, anti-rabbit: #474-1506, both in vitro model of GBM-derived cultures using both gain-of-function
KPL) were diluted 1:10,000 in blocking solution and incubated for 1 h at room tem-
as well as pathway inhibition model.
perature. Signal detection was performed on a film based system by applying Im-
mun-Star Western C Kit (Bio-Rad) or Super Signal West Pico Chemiluminescent
Substrate (Thermo Scientific).
3.2. Constitutive activation of the canonical WNT/b-catenin signaling
pathway with lentiviral construct encoding for mutant b-catenin
2.11. Cell proliferation assay
First, we examined our tumor cultures for intrinsic WNT/b-cate-
Mitochondrial activity for determining biological active cell mass was measured
nin pathway activity using quantitative PCR-based assessment of
using CellTiter 96Ò AQueous One Solution Cell Proliferation Assay according to the
instructions of the manufacturer (Promega GmbH). The individual values were re-
Axin2, DKK1 and CD44 expression compared to non-neoplastic hu-
trieved trough Elisa-based (Tecan) absorption measurements at 490 nm. The forma- man adult cortical tissue as control. This revealed variations in
zan dye generated by the metabolizing cells directly corresponds to viable cell mass WNT pathway activity among the cell lines, with cells showing
in culture. In each proliferation experiment we initially plated 20,000 cells per 24- low expression levels (i.e. U87: Axin2 0.05 and DKK1 1.7) as
well in 1 ml culture media. In order to plate as exact as possible the cells were pas-
well as many neurosphere cultures clearly over-expressing the
saged as described above directly prior plating.
investigated markers including, i.e. FF (Axin2 340, DKK1 24
and CD44 12) or NCH421k (Axin2 24, DKK1 1211 and CD44
2.12. In vitro motility assays 8) as shown in Fig. 2A.
To investigate the effects of WNT activation, we created glioma
Invasion and migration of GBM-derived cells was analyzed using a Boyden
Chamber assay. A 24-well transwell system (Corning Inc.) was used, with each well lines expressing the activating mutant form of b-catenin (S33Y
containing a permeable transwell insert containing a 6.4 mm polycarbonate mem- [27]). The presence of reporter gene GFP enabled the FACS-based
brane with 8 lm pores. To mimic three-dimensional extracellular environment and purification for cells overactivating the canonical b-catenin
investigate infiltrative potential of the cells, the inserts were coated with growth pathway.
factor reduced Matrigel (BD) and incubated for 1 h in at 37 °C. Migratory potential
of the cells were analyzed on 1% gelatine-coated inserts. Subsequently,
The activity of the pathway was measured using quantitative
7.5  105 GBM cells solved in 500 ll DMEM were placed on top of each insert mem- PCR-based evaluation of expression of previously reported tar-
brane. The bottom was filled with 500 ll DMEM media containing 10% FCS. After gets/negative feedback regulators of canonical WNT pathway,
incubation for 24–36 h, depending on the cell line. The upper side of the membrane DKK1 [8], and AXIN2 [23]. Axin2 was upregulated in all GBM-cul-
was then wiped carefully with a cotton swab to remove the rest of the plated cells.
tures expressing mutant form of b-catenin compared to control
The membrane was then fixed in 4% formaldehyde for 40 min and stained with
Haematoxylin. The migration/invasion of the GBM cells was evaluated by counting (Fig. 2B) highlighted by strongest upregulation the case of U87 cell
the cell nuclei on the lower side of the membrane under light microscope and line. This corresponded to intranuclear b-catenin localization in
20  magnification (6 view fields). immunostaining of cells transfected with mutant b-catenin
Cell numbers counted in the control condition (infection with cFUGW vector for (Fig. 2D).
the WNT overexpression, sh scramble for the knock down experiments, copGFP/
control for the DKK1 overexpression) were set as 100%. In relationship to that,
Overactivation of canonical WNT/b-catenin signaling led also to
counted cells with modulated WNT activity (S33Y, sh b-catenin, DKK1-overexpres- increase of DKK1 expression compared to control group (U87:
sion) were normalized to 100%. Axin2 501 and DKK1 7.8, NCH421k: Axin2 39.2 and DKK1
U.D. Kahlert et al. / Cancer Letters 325 (2012) 42–53 45

Fig. 1. Box plots representing TCGA expression data for WNT/b-catenin targets DKK1, DKK2, RUNX2, CD44, LEF1 and FZD1 in glioblastoma multiforme (GBM). Analysis of
different subgroups of GBM samples according to Verhaak et al. [56] (proneural (P), neural (N), classical (C) and mesenchymal (M)) showed WNT-target overexpression in the
mesenchymal subtype,  = dataset outliers, Student t-Test, # = p-value 6 0.002.

2.8, HSR-GBM1: Axin2 2 and DKK1 900, KK: Axin2 & DKK1 conditions. Vice versa, high basal levels of CD44 in U87 might ex-
about 2) (Fig. 2B). Furthermore, the mutant cells were character- plain why we could not detect its significant induction.
ized by moderate increased expression (2–3) of CD44, another Moreover, inhibition of intrinsic WNT/b-catenin pathway was
marker regulated by the WNT network [58]. Effective physiological achieved by lentiviral transfection with short hairpin (sh) RNAs
modulation of WNT targets Axin2 and DKK1 could be confirmed against b-catenin, leading to a decrease of Axin2 expression in all
with Western blot-based protein analysis (Fig. 2C). lines tested (Fig. 2E KK 1.3, FF 1.5, HSR-GBM1 over 2, GB
Upon WNT-overactivation, substantial induction of DKK1 in 2.8, Western blot exemplary shown in Fig. 2F for NCH421k and
case of HSR-GBM1 and Axin2 in U87 could presumably be ex- U87). Surprisingly, in this loss-of-function model we could not de-
plained by their explicit low expression levels in wild type (WT) tect any DKK1 transcripts any longer, except for U87, were we
46 U.D. Kahlert et al. / Cancer Letters 325 (2012) 42–53

Fig. 2. Modulation of WNT/b-catenin signaling pathway. Different basal levels of WNT/b-catenin pathway activity in wildtype GBM-derived cell cultures has been observed
as assessed by Axin2, DKK1 and CD44 expression normalized to non-neoplastic adult cortex (CTX) (A); transfection with mutant form of b-catenin (S33Y)-led to stable
overexpression of the WNT-targets (B) and increase of corresponding protein levels as exemplary shown for U87, KK and NCH421k cell lines (C); immunocytochemistry
against b-catenin in transfected lines revealed stronger nuclear accumulation as compared to control (cFUGW) as shown for U87S33Y (D); in contrary, inhibition of WNT/b-
catenin signaling via shRNA interference against b-catenin decreased expression of relevant markers (E) as well as protein levels of Axin2 (F), (A,B,E Student t-Test, p-
value 6 0.05).

proofed an over 2 decrease of DKK1 transcript (Fig. 2E). All tested sh RNA-based b-catenin inhibition and empty virus controls dis-
cultures showed also lower CD44 expression (1–2). played apparently similar growth dynamics (Fig. 3A and B).
Based on our experimental results we assume that DKK1 Interestingly, further characterizations of the genetically modi-
expression seen in the mesenchymal subgroup of GBMs correlates fied cells revealed significantly changed in vitro migratory poten-
with high WNT/b-catenin signaling and can be used as an indirect tial as tested in Boyden Chambers-based experiments. In our
marker for the canonical WNT pathway activity. investigations we used two different insert coatings- 1% gelatine
Taken together, we established robust systems enabling modu- and 1  Matrigel to test migration or invasiveness, respectively.
lation of the canonical WNT pathway and identified two targets, In comparison to empty virus control, all tested cell lines (n = 4)
Axin2 and DKK1, to monitor pathway activity in our subsequent showed stronger invasion (Fig. 3A: NCH421k 2, U87 54.3, KK
in vitro studies. 5.2, 276 7.3) as well as migration (NCH421k 2.8, U87 6, 1
and 1.2 for KK and 276) when infected with mutant b-catenin.
3.3. WNT/b-catenin level does not effect proliferation capacity but No significant difference was detected between wild type cell lines
directly correlates with in vitro cell motility or those infected with empty viral vector (data not shown).
Conversely, cells with reduced WNT/b-catenin signaling
To assess the effect of WNT/b-catenin modulation on prolifera- showed decreased numbers of invading and migrating cells as
tion of GBM cultures, we tested their metabolic activity using MTS compared to cells infected with empty virus control (Fig. 3B). In
following activation or inhibition of the pathway. In all tested cell fact, for KK only 50%, 80% for HSR-GBM1 and 48% for NCH421k
lines the analysis showed no significant differences in growth upon of the cells traversing the membrane, on both, gelainte- and Matri-
genetic pathway modulation. In fact, cells with mutated b-catenin/ gel surfaces. There was no significant motility difference compar-
U.D. Kahlert et al. / Cancer Letters 325 (2012) 42–53 47

Fig. 3. Modulation of WNT/b-catenin signaling influences in vitro motility of GBM-derived cultures. Activation of WNT/b-catenin pathway did not alter proliferation of
cultivated GBM tissue as assessed by MTS assay. WNT-pathway overactivation directly correlated with increased in vitro cell motility based on Boyden Chamber experiments
with gelatine and Matrigel coatings. There was no significant difference in motility or proliferation between WT and control lines (data not shown) (A). In contrast, WNT/b-
catenin inhibition resulted in significantly decreased cell motility (B), (A, B Student t-Test for qPCR, ANOVA with Newman–Keuls post hoc analysis for motility assays, p-
value 6 0.05).

ing WT cells and empty control cells except for KK showing 26% way activation, we analyzed mRNA expression levels of EMT-
less motility in both surfaces upon virus infection. Again, metabolic related transcription factors ZEB1, Twist, Snail and Slug. Indeed,
activity assessed by MTS was not altered upon WNT/b-catenin increased levels of ZEB1 and Twist (both 2–3.5) could be de-
inhibition (Fig. 3B). tected. KK showed the strongest induction of Snail (3.7). These
alterations could be proved also on protein level as shown in
3.4. WNT activation induces expression of EMT activators Fig. 4A. Another important feature of EMT is an increased ratio
of N-Cadherin to E-Cadherin. The expression of E-Cadherin could
Increased invasive potential of cancer cells in solid tumors has not be observed in any of the GBM cultures. Nevertheless N-Cad-
been associated with EMT of the migrating tumor cell population herin transcript levels were clearly enhanced in cells with acti-
[55,57]. To determine if induction of EMT in GBM-derived cell cul- vated WNT/b signaling (up to 3 for NCH421k and over 2 for
tures may explain their increased invasion following WNT path- U87, Fig. 4A).
48 U.D. Kahlert et al. / Cancer Letters 325 (2012) 42–53

Fig. 4. Activation of WNT/b-catenin signaling promotes the expression of EMT activators and CD133. S33Y transfected cell cultures show induction of epithelial-to-
mesenchymal (EMT) activators ZEB1, Twist, Snail as well as N-Cadherin as showed at mRNA and protein levels (A), whereas the inhibition of intrinsic WNT/b-catenin
signaling led to their decreased transcription and translation (B). Also a putative brain tumor stem cell marker CD133 appeared to up- and down regulated in response to
WNT/b-catenin pathway modulation with the highest transcript and protein level following WNT activation (C,D) (A,B Student t-Test, p-value 6 0.05).

Consistent with this finding, shRNA interference against b-cate- of the canonical WNT/b-catenin cascade, we observed overexpres-
nin reduced the mRNA levels of EMT factors (2 decrease for Twist sion of Prominin1 mRNA (3 for NCH421k, 2 for NCH644, KK and
in GB, 3 ZEB1 and Twist for FF and up to 2.7 for N-Cadherin in HSR-GBM1) as well as increase of corresponding protein (Fig. 4C).
HSR-GBM1; Fig. 4B). Furthermore, clear reduction of ZEB1 and Similarly, following inhibition of WNT/b-catenin pathway we
Twist by Western blot could be demonstrated. consistently observed decreased signals for CD133 (1.5–3 for
U87, HSR-GBM1 and KK, Fig. 4D).

3.5. Activation of WNT/b-catenin pathway leads to the increase of


mRNA and protein levels of putative brain tumor stem cell marker
CD133 3.6. Overexpression of DKK1 did not enhance migratory behavior

Next, we evaluated the relationship between levels of WNT DKK1 has been previously reported to act in a feedback loop for
pathway activity, expression of mesenchymal activators and brain canonical WNT/b-catenin pathway activation [8]. Based on the
tumor stem cell marker CD133 (Prominin1). Following activation overexpression of DKK1 in the mesenchymal subgroup of GBMs
U.D. Kahlert et al. / Cancer Letters 325 (2012) 42–53 49

(TCGA data), we tested weather a DKK1 overexpression has any di- 4. Discussion
rect effect on in vitro motility.
We established a lentiviral system to overexpress DKK1 and Aberrant activation of developmental signaling pathways
validated its efficiency using qPCR and Western blot (Fig. 5A). Next, important in stem cells are hallmarks of neoplastic transformation
we evaluated the effect of DKK1 overexpression of migration and in the brain and various other tissues (i.e. Hedgehog [5], Hippo [12]
invasion of glioblastoma using the Boyden Chamber assay. Inter- and Notch [16]). Our current work aims at deciphering the role of
estingly, these assays revealed a diminished motility of cells over- the canonical WNT/b-catenin pathway in the biology of adult glial
expressing DKK1 in both conditions (percentages traversing cells tumors. We present clinical data suggesting an important role for
upon DKK1 overexpression comparing to control, on gelatine: the canonical WNT/b-catenin signaling network in a subset of hu-
U87 2%, 276 13%; on Matrigel: U87 10%, 276 39%, no difference be- man-derived GBM samples, as well as an in vitro experimental
tween WT and control [data not shown], Fig. 5C). model of pathway modulation supporting its relevance especially
in the regulation of tumor cell migration and invasion.
The analysis of the TCGA-derived gene expression data of over
3.7. Nuclear localization of b-catenin is predominantly found in the 200 tumor samples enabled the identification of several subclasses
infiltration zone in patient-derived glioblastoma specimens of glioblastomas [38]. Interestingly, the gene expression data pub-
lished as a supplementary material by Verhaak et al. [56] showed
Immunohistochemical assessment of GBM specimens, subdi- up to 6 up-regulation of Dickkopf1 (DKK1), 3 for Frizzled1
vided in tumor parenchyma and infiltration zone (Fig. 6A) upon (FZD1) and Runx2 almost exclusively in the mesenchymal group
exclusion of non-tumorous CD31 positive endothelial cells of GBMs, characterized by frequent multi-focal growth and early
(Fig. 6B), revealed cells with strong intranuclear b-catenin signal tumor recurrence leading to fast clinical deterioration and nega-
in all 30 observed cases (Fig. 6C). The quantity ranged from few tively influencing of patient overall survival. These findings at-
cells to a substantial fraction of positive nuclei (over 5%) in 14 from tracted our attention to a potential role of canonical WNT
30 samples (Fig. S2). Interestingly, in 11 GBM-derived samples we signaling in adult gliomas. Since DKK1, a direct target of b-catenin
observed higher numbers of cells with intranuclear b-catenin sig- [37], playing a pivotal role in the negative feedback-based control
nal in peripheral infiltration zone of the tumor as compared to of pathway activity [8], showed the highest expression in mesen-
its central parenchyma (for evaluation of average pathological chymal GBMs and in our in vitro experiments was not able to in-
scoring of the 30 samples see Fig. 6D). In 4 tumors WNT activation, duce the motility of cultivated GBM cells, we hypothesize, that
as evidenced by nuclear b-catenin, was seen exclusively in the elevated DKK1 expression levels indirectly indicate WNT pathway
infiltration zone (Fig. S2). overactivation, responsible for observed phenotypic changes. In
A high frequency of co-localization of the endothelial marker fact, secreted soluble frizzled receptors 1 and 2 were predomi-
CD31 and nuclear b-catenin (data not published) was also ob- nantly inactivated in primary GBMs [20].
served, consistent with previous reports [61]. Aside from the vas- Immunohistochemical assessment of 30 patient-derived GBM
cular staining, intranuclear b-catenin signal was significantly samples, consisting of tumor parenchyma, infiltration zone and
lower in non-cancerous sections compared to the tumor tissue non-cancerous adjacent brain tissue showed significantly more
(Fig. 6D).

Fig. 5. Biological effects of WNT inhibitor overexpression – DKK1. Using 3rd generation lentiviral transfection system DKK1 was stably overexpressed in GBM-derived
cultures as shown with transcript and corresponding protein (A). It resulted in diminished in vitro cell motility on gelatine- and Matrigel coated Boyden Chambers (B) (A,
Student t-Test, B, ANOVA with Newman-Keuls post hoc analysis for motility assays, p-value 6 0.05).
50 U.D. Kahlert et al. / Cancer Letters 325 (2012) 42–53

Fig. 6. Presence of cells with intranuclear b-catenin signal in IHC assessment of patient-derived GBM samples. In 11 of 30 GBM samples increased number of b-catenin
positive nuclei at the tumors infiltration zone as compared to tumor parenchyma (examples shown by red arrows) could be observed. Overall averaged scoring revealed
significant increased numbers of cells with nuclear b-catenin in the infiltration zone compared to the parenchyma. The lowest levels of intranuclear b-catenin signal were
observed in the adjacent neuropil (D, ANOVA p-value 6 0.05); H&E (A), endothelial staining CD31 (B) and intranuclear b-catenin (C) of patient-derived GBM tissue. (For
interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

cells displaying intranuclear b-catenin signal within the tumor an overexpression of Dishevelled-2 (DVL2), a key component of the
parenchyma and infiltration zone as compared to adjacent brain. WNT pathway in patient-derived samples of GBM [43]. Interest-
Interestingly, the strongest signal was present within tumor infil- ingly, RNA-interference mediated inhibition of Dvl2 expression
tration zones adjacent to healthy tissue (Fig. 6C + D). These find- blocked in vitro proliferation of tumor-derived cells, promoting
ings suggest that, similar to colorectal carcinoma [7], in a their differentiation as well as inhibiting tumor formation capacity
subgroup of GBMs the activity of the canonical WNT/b-catenin sig- upon intracranial implantation. Additionally, evaluation of clinical
naling pathway is directly associated with increased motility of the outcome of patients with GBM revealed b-catenin as a negative
tumor cells leading to expansion of the malignant process. Further prognostic factor for overall survival [45].
characterization of cells with intranuclear b-catenin signal within In the current study we found that transduction of a mutant
tumor infiltration zone is compulsory. form of b-catenin (S33Y) leading to constitutive activity of the
The possible role of canonical WNT signaling in malignant glial pathway significantly increased in vitro invasion and migration in
tumors has already been reported by Pulvirenti et al., who showed all studied cell lines with no effect on their growth kinetic. Inter-
U.D. Kahlert et al. / Cancer Letters 325 (2012) 42–53 51

estingly, previously reported knock down experiments against differences these interesting data cannot be directly compared
WNT2 and b-catenin showed significant impairment of the prolif- with our observations. Here we show that the overexpression of
eration capacity of GBM cultures both in vitro and in vivo [42]. Con- EMT activators in GBM-derived cell lines occurs following the
cordantly, blocking the secretion of multiple WNTs by silencing of induction of WNT/b-catenin pathway activity. Amongst the pa-
Evi resulted in diminished proliferation capacity of GBM cultures tient-derived samples the intranuclear b-catenin staining could
[3]. These observations could not be reproduced in our RNA-inter- be rarely observed in P5% of all tumor cells. Therefore the thresh-
ference mediated loss-of-function model. In fact, inhibition of the old set by Nagaishi et al. at P10% of total cells may led to under-
intrinsic canonical WNT pathway activity had no pronounced ef- estimation of the possible GBM samples showing WNT-
fect on the growth as determined by MTS assay. This might be dependent overexpression of EMT activators. Furthermore, the lack
due to the significance of non-canonical WNT signaling mediated of b-catenin staining an exclusive assessment of the tumor paren-
by both Evi [1] and Dvl2 [43] on the glioma cell division, not mod- chyma without infiltration zone precludes any direct comparison
ulated in our experiments. with our study and does not exclude the presence of subpopula-
Transfection with the lentiviral construct encoding mutated b- tions of cells undergoing EMT-like changes in some tumor
catenin promoted the induction of Jag1 mRNA expression (Supple- compartments.
mentary Fig. S4), possibly linking WNT to the Notch pathway. Sim- EMT has also been associated with increased resistance to radi-
ilar interactions have been reported in the developing human brain ation and chemotherapy in several solid tumors [52]. Furthermore,
[28]. Furthermore, cross-talk between Notch and Sonic Hedgehog WNT/b-catenin signaling has also been implicated in promoting
pathways in GBM cultures have also recently been described resistance against classical treatment modalities, attributed to tu-
[50]. However, further studies addressing the involvement of mor stem-like cell population in various cancer types including
Notch-signaling upon WNT pathway activation are necessary. GBM [4]. Interestingly, following WNT activation, GBM cells
The WNT network is also involved in the regulation of stem cell showed increased mRNA transcripts of ATP-binding cassette
pool during development as shown by numerous reports (for re- (ABC) transporter ABCG2, involved in resistance against chemo-
view see [10]). It has also been reported as one of the regulatory therapy in GBMs [6] (Supplementary Fig. S6).
mechanism of stem-like cell population in a number of malignan- Taken together, our results point out for the first time, the
cies including, i.e. colon and breast cancer or acute myeloid lym- unequivocal significance of WNT/b-catenin-mediated EMT induc-
phoma [22,44]. Here we show that manipulation of WNT activity tion in regulation of the invasiveness of human malignant gliomas
influenced the expression of Prominin-1 indicating its possible role and increases the signal for CD133 on mRNA and protein level sug-
in the regulation of brain tumor stem-like cells (BTSCs). Similar to gesting its involvement in regulating BTSC-pool, addressed par-
our findings, an increase of CD133 + cells in glioma cultures medi- tially by recent reports [24,33]. Moreover, the spatial distribution
ated by the direct activator of the WNT/b-catenin pathway PLAGL2 of the nuclear b-catenin signal in patient-derived samples, found
has already been reported [64]. predominantly within the invasive front of the tumor, suggests
In our experiments, an over-activation of the canonical WNT/b- the pivotal role of this pathway in regulation of in vivo malignant
catenin signaling constantly led to an increased in vitro motility. In cell motility. These findings may contribute to the development
other solid tumors (i.e. bladder [59], breast [47], pancreas [25]) in- of an alternative therapeutic strategy based on the interference
creased invasiveness is often mediated by complete genetic repro- with the canonical WNT/b-catenin pathway using numerous effi-
gramming of the cancer cell- a process termed epithelial-to- cient small molecules (i.e. [18]) specifically targeting the migrating
mesenchymal transition (EMT). Furthermore, this phenomenon cell population and thus preventing tumor recurrence.
plays a pivotal role not only in tumor dissemination but also medi-
ates the resistance to treatment modalities, escape from the im-
Acknowledgements
mune surveillance, as well as stem cell maintenance.
Predominantly occurring within the tumor infiltration zone, EMT
We gratefully thank K. Geiger, Core Facility 1, Department of
can be induced by several factors including cytokines, hypoxia or
Internal Medicine, University Medical Center Freiburg, Germany
cellular and extracellular matrix [55,57]. Interestingly, highly con-
for the flow-cytometry sorting and C. Haas, Experimental Epilepsy,
served canonical WNT/b-catenin pathway is one of the known
Medical Center Freiburg for the generous opportunity to use the
inducers of EMT in both physiological and pathological conditions
qPCR system in her lab.
(i.e. [26,49]). Similar to these reports, over-activation of the WNT
This work was supported by the Comprehensive Cancer Center
pathway in GBM cultures induced the expression of EMT activators
Freiburg Seeding Grant (Deutsche Krebshilfe) awarded to J. Mac-
ZEB1, Snail, Twist, Slug as well as N-Cadherin being responsible for
iaczyk and G. Nikkhah.
the alteration of the cell-cell contact and induction of increased cell
motility observed in the current study. Interestingly Snail has al-
ready been shown to directly modulate in vitro motility of GBM- Appendix A. Supplementary data
derived cells [21]. Furthermore, similar to silencing of the intrinsic
WNT/b-catenin activity via RNA-interference, overexpression of Supplementary data associated with this article can be found, in
the WNT pathway inhibitor DKK1 led to decreased mRNA expres- the online version, at http://dx.doi.org/10.1016/
sion of EMT-related transcription factors (Supplementary Fig. S5) j.canlet.2012.05.024.
and abrogation of the in vitro migratory/invasive capacity of the
cells. Similarly, in prostate cancer, inhibition of WNT signaling
References
leads to the reversal of EMT [63]. The presence and possible role
of the EMT-like process in glioblastoma is currently very contro- [1] T. Adell, E. Salò, M. Boutros, K. Bartscherer, Smed-Evi/Wntless is required for
versial. Very recent report by Cheng et al. [9] proved the overex- beta-catenin-dependent and -independent processes during planarian
regeneration, Development 136 (2009) 905–910.
pression of EMT activators in a subgroup of GBMs showing
[2] M. Antonelli, F.R. Buttarelli, A. Arcella, S. Nobusawa, V. Donofrio, H. Oghaki,
unequivocal negative prognostic value of the mesenchymal transi- et al., Prognostic significance of histological grading, p53 status, YKL-40
tion metagen for patient survival. In contrast, Nagaishi et al. [35] expression, and IDH1 mutations in pediatric high-grade gliomas, J. Neurooncol.
reported on the presence of EMT activators in gliosarcomas but 99 (2010) 209–215.
[3] I. Augustin, V. Goidts, A. Bongers, G. Kerr, G. Vollert, B. Radlwimmer, et al., The
not in glioblastomas based on immunohistochemical analysis of Wnt secretion protein Evi/Gpr177 promotes glioma tumourigenesis, EMBO
patient-derived tumor specimens. However, due to methodological Mol. Med. 4 (2012) 38–51.
52 U.D. Kahlert et al. / Cancer Letters 325 (2012) 42–53

[4] S. Bao, Q. Wu, R.E. McLendon, Y. Hao, Q. Shi, A.B. Hjelmeland, et al., Glioma term expanded fetal human neural precursor cells after transplantation into
stem cells promote radioresistance by preferential activation of the DNA the adult rat brain, Stem Cells Dev. 18 (2009) 1043–1058.
damage response, Nature 444 (2006) 756–760. [33] S.A. Mikheeva, A.M. Mikheev, A. Petit, R. Beyer, R.G. Oxford, L. Khorasani, et al.,
[5] E.E. Bar, A. Chaudhry, A. Lin, X. Fan, K. Schreck, W. Matsui, et al., Cyclopamine- TWIST1 promotes invasion through mesenchymal change in human
mediated hedgehog pathway inhibition depletes stem-like cancer cells in glioblastoma, Mol. Cancer 9 (2010) 194.
glioblastoma, Stem Cells 25 (2007) 2524–2533. [34] P.J. Morin, A.B. Sparks, V. Korinek, N. Barker, H. Clevers, B. Vogelstein, et al.,
[6] A.-M. Bleau, J.T. Huse, E.C. Holland, The ABCG2 resistance network of Activation of beta-catenin-Tcf signaling in colon cancer by mutations in beta-
glioblastoma, Cell Cycle 8 (2009) 2936–2944. catenin or APC, Science 275 (1997) 1787–1790.
[7] T. Brabletz, A. Jung, K. Hermann, K. Günther, W. Hohenberger, T. Kirchner, [35] M. Nagaishi, W. Paulus, B. Brokinkel, A. Vital, Y. Tanaka, Y. Nakazato, et al.,
Nuclear overexpression of the oncoprotein beta-catenin in colorectal cancer is Transcriptional Factors for Epithelial-Mesenchymal Transition are Associated
localized predominantly at the invasion front, Pathol. Res. Pract. 194 (1998) with Mesenchymal Differentiation in Gliosarcoma, Brain Pathology, Zurich,
701–704. Switzerland, 2012.
[8] M.N. Chamorro, D.R. Schwartz, A. Vonica, A.H. Brivanlou, K.R. Cho, H.E. Varmus, [36] J.T. Nauseef, M.D. Henry, Epithelial-to-mesenchymal transition in prostate
FGF-20 and DKK1 are transcriptional targets of beta-catenin and FGF-20 is cancer: paradigm or puzzle?, Nat Rev. Urol. 8 (2011) 428–439.
implicated in cancer and development, EMBO J. 24 (2005) 73–84. [37] A. Niida, T. Hiroko, M. Kasai, Y. Furukawa, Y. Nakamura, Y. Suzuki, et al., DKK1,
[9] W.-Y. Cheng, J.J. Kandel, D.J. Yamashiro, P. Canoll, D. Anastassiou, A multi- a negative regulator of Wnt signaling, is a target of the beta-catenin/TCF
cancer mesenchymal transition gene expression signature is associated with pathway, Oncogene 23 (2004) 8520–8526.
prolonged time to recurrence in glioblastoma, PLoS ONE 7 (2012) e34705. [38] H. Noushmehr, D.J. Weisenberger, K. Diefes, H.S. Phillips, K. Pujara, B.P.
[10] H. Clevers, Wnt/beta-catenin signaling in development and disease, Cell 127 Berman, et al., Identification of a CpG island methylator phenotype that
(2006) 469–480. defines a distinct subgroup of glioma, Cancer Cell 17 (2010) 510–522.
[11] L.A.D. Cooper, J. Kong, D.A. Gutman, F. Wang, J. Gao, C. Appin, et al., Integrated [39] R. Nusse, Wnt signaling and stem cell control, Cell Res. 18 (2008) 523–527.
morphologic analysis for the identification and characterization of disease [40] B.A. Orr, H. Bai, Y. Odia, D. Jain, R.A. Anders, C.G. Eberhart, Yes-associated
subtypes, J. Am. Med. Inform. Assoc. 19 (2012) 317–323. protein 1 is widely expressed in human brain tumors and promotes
[12] M. Cordenonsi, F. Zanconato, L. Azzolin, M. Forcato, A. Rosato, C. Frasson, et al., glioblastoma growth, J. Neuropathol. Exp. Neurol. 70 (2011) 568–577.
The hippo transducer TAZ confers cancer stem cell-related traits on breast [41] H.S. Phillips, S. Kharbanda, R. Chen, W.F. Forrest, R.H. Soriano, T.D. Wu, et al.,
cancer cells, Cell 147 (2011) 759–772. Molecular subclasses of high-grade glioma predict prognosis, delineate a
[13] T.A. DiMeo, K. Anderson, P. Phadke, C. Fan, C. Feng, C.M. Perou, et al., A novel pattern of disease progression, and resemble stages in neurogenesis, Cancer
lung metastasis signature links Wnt signaling with cancer cell self-renewal Cell 9 (2006) 157–173.
and epithelial-mesenchymal transition in basal-like breast cancer, Cancer Res. [42] P. Pu, Z. Zhang, C. Kang, R. Jiang, Z. Jia, G. Wang, et al., Downregulation of Wnt2
69 (2009) 5364–5373. and beta-catenin by siRNA suppresses malignant glioma cell growth, Cancer
[14] Y.-F. Dong, D.Y. Soung, E.M. Schwarz, R.J. O’Keefe, H. Drissi, Wnt induction of Gene Ther. 16 (2009) 351–361.
chondrocyte hypertrophy through the Runx2 transcription factor, J. Cell. [43] T. Pulvirenti, M. Van Der Heijden, L.A. Droms, J.T. Huse, V. Tabar, A. Hall,
Physiol. 208 (2006) 77–86. Dishevelled 2 signaling promotes self-renewal and tumorigenicity in human
[15] T. Dull, R. Zufferey, M. Kelly, R.J. Mandel, M. Nguyen, D. Trono, et al., A third- gliomas, Cancer Res. 10 (2011).
generation lentivirus vector with a conditional packaging system, J. Virol. 72 [44] T. Reya, H. Clevers, Wnt signalling in stem cells and cancer, Nature 434 (2005)
(1998) 8463–8471. 843–850.
[16] X. Fan, L. Khaki, T.S. Zhu, M.E. Soules, C.E. Talsma, N. Gul, et al., NOTCH [45] M. Rossi, L. Magnoni, C. Miracco, E. Mori, P. Tosi, L. Pirtoli, et al., b-catenin and
pathway blockade depletes CD133-positive glioblastoma cells and inhibits Gli1 are prognostic markers in glioblastoma, Cancer Biol. Ther. 11 (2011) 753–
growth of tumor neurospheres and xenografts, Stem Cells 28 (2010) 5–16. 761.
[17] R. Firestein, A.J. Bass, S.Y. Kim, I.F. Dunn, S.J. Silver, I. Guney, et al., CDK8 is a [46] D.D. Sarbassov, D.A. Guertin, S.M. Ali, D.M. Sabatini, Phosphorylation and
colorectal cancer oncogene that regulates beta-catenin activity, Nature 455 regulation of Akt/PKB by the rictor-mTOR complex, Science 307 (2005) 1098–
(2008) 547–551. 1101.
[18] N. Fujii, L. You, Z. Xu, K. Uematsu, J. Shan, B. He, et al., An antagonist of [47] D. Sarrio, C.K. Franklin, A. Mackay, J.S. Reis-Filho, C.M. Isacke, Epithelial and
dishevelled protein–protein interaction suppresses beta-catenin-dependent mesenchymal subpopulations within normal basal breast cell lines exhibit
tumor cell growth, Cancer Res. 67 (2007) 573–579. distinct stem cell/progenitor properties, Stem Cells (2011).
[19] R. Galli, E. Binda, U. Orfanelli, B. Cipelletti, A. Gritti, S. De Vitis, et al., Isolation [48] L. Sastry, T. Johnson, M.J. Hobson, B. Smucker, K. Cornetta, Titering lentiviral
and characterization of tumorigenic, stem-like neural precursors from human vectors: comparison of DNA, RNA and marker expression methods, Gene Ther.
glioblastoma, Cancer Res. 64 (2004) 7011–7021. 9 (2002) 1155–1162.
[20] S. Götze, M. Wolter, G. Reifenberger, O. Müller, S. Sievers, Frequent promoter [49] C. Scheel, E.N. Eaton, S.H.-J. Li, C.L. Chaffer, F. Reinhardt, K.-J. Kah, et al.,
hypermethylation of Wnt pathway inhibitor genes in malignant astrocytic Paracrine and autocrine signals induce and maintain mesenchymal and stem
gliomas, Int. J. Cancer 126 (2010) 2584–2593. cell states in the breast, Cell 145 (2011) 926–940.
[21] S.-P. Han, J.-H. Kim, M.-E. Han, H.-E. Sim, K.-S. Kim, S. Yoon, et al., SNAI1 is [50] K.C. Schreck, P. Taylor, L. Marchionni, V. Gopalakrishnan, E.E. Bar, N. Gaiano,
involved in the proliferation and migration of glioblastoma cells, Cell. Mol. et al., The Notch target Hes1 directly modulates Gli1 expression and Hedgehog
Neurobiol. 31 (2011) 489–496. signaling: a potential mechanism of therapeutic resistance, Clin. Cancer Res.
[22] J. Izrailit, M. Reedijk, Developmental pathways in breast cancer and breast 16 (2010) 6060–6070.
tumor-initiating cells: therapeutic implications, Cancer Lett. (2011). [51] S.K. Singh, C. Hawkins, I.D. Clarke, J.A. Squire, J. Bayani, T. Hide, et al.,
[23] E. Jho, T. Zhang, C. Domon, C.-K. Joo, J.-N. Freund, F. Costantini, Wnt/beta- Identification of human brain tumour initiating cells, Nature 432 (2004) 396–
catenin/Tcf signaling induces the transcription of Axin2, a negative regulator 401.
of the signaling pathway, Mol. Cell. Biol. 22 (2002) 1172–1183. [52] M.A. Smit, D.S. Peeper, Zeb1 is required for TrkB-induced epithelial-
[24] X. Jin, H.-Y. Jeon, K.M. Joo, J.-K. Kim, J. Jin, S.H. Kim, et al., Frizzled 4 regulates mesenchymal transition, anoikis resistance and metastasis, Oncogene 30
stemness and invasiveness of migrating glioma cells established by serial (2011) 3735–3744.
intracranial transplantation, Cancer Res. 71 (2011) 3066–3075. [53] R. Stupp, M.E. Hegi, W.P. Mason, M.J. van den Bent, M.J.B. Taphoorn, R.C. Janzer,
[25] S. Joost, L.L. Almada, V. Rohnalter, P.S. Holz, A.M. Vrabel, M.G. Fernandez- et al., Effects of radiotherapy with concomitant and adjuvant temozolomide
Barrena, et al., GLI1 inhibition promotes epithelial-to-mesenchymal transition versus radiotherapy alone on survival in glioblastoma in a randomised phase
in pancreatic cancer cells, Cancer Res. (2011). III study: 5-year analysis of the EORTC-NCIC trial, Lancet. Oncol. 10 (2009)
[26] R. Kemler, A. Hierholzer, B. Kanzler, S. Kuppig, K. Hansen, M.M. Taketo, et al., 459–466.
Stabilization of beta-catenin in the mouse zygote leads to premature [54] G. Tabatabai, R. Stupp, M.J. van den Bent, M.E. Hegi, J.C. Tonn, W. Wick, et al.,
epithelial-mesenchymal transition in the epiblast, Development 131 (2004) Molecular diagnostics of gliomas: the clinical perspective, Acta Neuropathol.
5817–5824. 120 (2010) 585–592.
[27] F.T. Kolligs, G. Hu, C.V. Dang, E.R. Fearon, Neoplastic transformation of RK3E by [55] J.P. Thiery, H. Acloque, R.Y.J. Huang, M.A. Nieto, Epithelial-mesenchymal
mutant beta-catenin requires deregulation of Tcf/Lef transcription but not transitions in development and disease, Cell 139 (2009) 871–890.
activation of c-myc expression, Mol. Cell. Biol. 19 (1999) 5696–5706. [56] R.G.W. Verhaak, K.A. Hoadley, E. Purdom, V. Wang, Y. Qi, M.D. Wilkerson, et al.,
[28] C. Kwon, P. Cheng, I.N. King, P. Andersen, L. Shenje, V. Nigam, et al., Notch post- Integrated genomic analysis identifies clinically relevant subtypes of
translationally regulates b-catenin protein in stem and progenitor cells, Nat. glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and
Cell Biol. 13 (2011) 1244–1251. NF1, Cancer Cell 17 (2010) 98–110.
[29] T. Liu, A.J. DeCostanzo, X. Liu, Hy Wang, S. Hallagan, R.T. Moon, et al., G protein [57] U. Wellner, J. Schubert, U.C. Burk, O. Schmalhofer, F. Zhu, A. Sonntag, et al., The
signaling from activated rat frizzled-1 to the beta-catenin-Lef-Tcf pathway, EMT-activator ZEB1 promotes tumorigenicity by repressing stemness-
Science 292 (2001) 1718–1722. inhibiting microRNAs, Nat. Cell Biol. 11 (2009) 1487–1495.
[30] C. Lois, E.J. Hong, S. Pease, E.J. Brown, D. Baltimore, Germline transmission and [58] V.J. Wielenga, R. Smits, V. Korinek, L. Smit, M. Kielman, R. Fodde, et al.,
tissue-specific expression of transgenes delivered by lentiviral vectors, Science Expression of CD44 in Apc and Tcf mutant mice implies regulation by the WNT
295 (2002) 868–872. pathway, Am. J. Pathol. 154 (1999) 515–523.
[31] D.N. Louis, H. Ohgaki, O.D. Wiestler, W.K. Cavenee, P.C. Burger, A. Jouvet, The [59] Y. Xue, L. Li, D. Zhang, K. Wu, Y. Chen, J. Zeng, et al., Twisted epithelial-to-
2007 WHO classification of tumours of the central nervous system, Acta mesenchymal transition promotes progression of surviving bladder cancer T24
Neuropathol. 114 (2007) 97–109. cells with hTERT-dysfunction, PLoS ONE 6 (2011) e27748.
[32] J. Maciaczyk, I. Singec, D. Maciaczyk, A. Klein, G. Nikkhah, Restricted [60] H.W. Yang, L.G. Menon, P.M. Black, R.S. Carroll, M.D. Johnson, SNAI2/Slug
spontaneous in vitro differentiation and region-specific migration of long- promotes growth and invasion in human gliomas, BMC Cancer 10 (2010) 301.
U.D. Kahlert et al. / Cancer Letters 325 (2012) 42–53 53

[61] H. Yano, A. Hara, K. Takenaka, K. Nakatani, J. Shinoda, K. Shimokawa, et al., [64] H. Zheng, H. Ying, R. Wiedemeyer, H. Yan, S.N. Quayle, E.V. Ivanova, et al.,
Differential expression of beta-catenin in human glioblastoma multiforme and PLAGL2 regulates Wnt signaling to impede differentiation in neural stem cells
normal brain tissue, Neurol. Res. 22 (2000) 650–656. and gliomas, Cancer Cell 17 (2010) 497–509.
[62] N. Zhang, P. Wei, A. Gong, W.-T. Chiu, H.-T. Lee, H. Colman, et al., FoxM1 [65] I. Zlobec, A. Lugli, Epithelial mesenchymal transition and tumor budding in
promotes b-catenin nuclear localization and controls Wnt target-gene aggressive colorectal cancer: tumor budding as oncotarget, Oncotarget 1
expression and glioma tumorigenesis, Cancer Cell 20 (2011) 427–442. (2010) 651–661.
[63] J.-H. Zhao, Y. Luo, Y.-G. Jiang, D.-L. He, C.-T. Wu, Knockdown of b-catenin
through shRNA cause a reversal of EMT and metastatic phenotypes induced by
HIF-1a, Cancer Invest. 29 (2011) 377–382.

You might also like