Neuro-Oncology

Neuro-Oncology
Neuro-Oncology 18(11), 14871497, 2016
doi:10.1093/neuonc/now053
Neuro-Oncology 2016; 0, 1 11, doi:10.1093/neuonc/now053
Advance Access date 3 April 2016

Insulin-like growth factor binding protein-2 regulates b-catenin

signaling pathway in glioma cells and contributes to poor to poor
together contributes
patient prognosis

Downloaded from http://neuro-oncology.oxfordjournals.org/ at Mihai Eminescu Central University Library of Iasi on November 25, 2016
Shilpa S. Patil, Priyanka Gokulnath, Mohsin Bashir, Shivayogi D. Shwetha, Janhvi Jaiswal, Arun H. Shastry,
Arivazhagan Arimappamagan, Vani Santosh, and Paturu Kondaiah

Molecular Reproduction, Development and Genetics department, Indian Institute of Science, Bangalore, India (S.S.P., P.G., M.B., P.K.);
Department of Neuropathology, National Institute of Mental Health and Neuro Sciences, Bangalore, India (S.D.S., J.J., V.S.); Clinical
Neurosciences, National Institute of Mental Health and Neuro Sciences, Bangalore, India (A.H.S.); Neurosurgery, National Institute of
Mental Health and Neuro Sciences, Bangalore, India (A.A.)

Corresponding Author: Paturu Kondaiah, PhD, Department of Molecular Reproduction Development and Genetics, Indian Institute of Science,
Bangalore 560 012 India (paturu@mrdg.iisc.ernet.in).

Background. Upregulation of insulin-like growth factor binding protein 2 (IGFBP-2) is often associated with aggressiveness of glio-
blastoma (GBM) and contributes to poor prognosis for GBM patients. In view of the regulation of b-catenin by IGFBP-2 in breast
cancer and the crucial role of b-catenin pathway in glioma invasion, proliferation and maintenance of glioma stem cells, the
mechanism of regulation of b-catenin by IGFBP-2, and its role in GBM prognosis was studied.
Methods. Regulation of the b-catenin pathway was studied by immunocytochemistry, Western blot analysis, luciferase assays,
and real-time RT-PCR. The role of IGFBP-2 was studied by subcutaneous tumor xenografts in immunocompromised mice using
glioma cells engineered to express IGFBP-2 and its domains. GBM patient tumor tissues (n 112) were analyzed for expression of
IGFBP-2 and b-catenin by immunohistochemistry. Survival analysis was performed employing Cox regression and Kaplan-Meier
survival analyses.
Results. IGFBP-2 knockdown in U251, T98G, and U373 or overexpression in LN229 and U87 cells revealed a role for IGFBP-2 in
stabilization of b-catenin and regulation of its nuclear functions involving integrin-mediated inactivation of GSK3b. Similar results
were obtained upon overexpression of the C-terminal domain of IGFBP-2 but not the N-terminal domain. Subcutaneous xenograft
tumors overexpressing either full-length or the C-terminal domain of IGFBP-2 showed larger volume as compared with controls.
Coexpression of high levels of IGFBP-2 and b-catenin was associated with worse prognosis (P .001) in GBM patients.
Conclusion. IGFBP-2 potentiates GBM tumor growth by the activation of the b-catenin pathway through its C-terminal domain,
and their coexpression possibly contributes to worse patient prognosis.

Keywords: GBM prognosis, GSK3b, IGFBP-2 C-domain, b-catenin signaling.

Glioma accounts for almost 30% of all central nervous system
tumors and 80% of primary malignant tumors of the central
nervous system. Glioblastoma (GBM) comprises 45% of all glio-
mas and is associated with the poorest survival.1
Insulin-like growth factor binding protein-2 (IGFBP-2) is con-
ventionally known as the modulator of IGF signaling.2 Binding
of IGFBP to IGF regulates either the bioavailability to the receptor
or the half-life of IGF in circulation.2 Upregulation of IGFBP-2 is
often associated with aggressiveness of GBM and poor patient
prognosis.3 IGFBP-2 overexpression in glioma cell lines has been
shown to increase migration, invasion, and chemoresistance.4 7
Moreover, a single chain monoclonal antibody, which we devel-
oped against IGFBP-2, inhibited invasiveness of GBM cells.8
Although most of the IGF-independent functions of IGFBP-2
are known to occur via integrin signaling,4,6,7 the underlying
pathways and exact mechanisms behind their regulation re-
main unknown. So far in glioma, it has been shown that
IGFBP-2 activates NFkB9 and JNK5 pathways via a5b1 integrin
signaling, which is responsible for IGFBP-2-mediated survival
and invasion.

Received 4 August 2015; accepted 6 March 2016

# The Author(s) 2016. Published by Oxford University Press on behalf of the Society for Neuro-Oncology. All rights reserved.
For permissions, please e-mail: journals.permissions@oup.com.

1487
1 of 11
Patil et al.: IGFBP-2 regulates
regulates b-catenin
b-catenin in
in GBM
GBM
Previously, we have reported the stabilization of b-catenin
by IGFBP-2 in breast cancer cells and its association with
lymph node metastasis.10 However, the exact mechanism un-
derlying this regulation and its significance with respect to GBM
was unclear. The b-catenin pathway is known to play a crucial
role in glioma cell proliferation,11 invasion,12 and maintenance
of glioma-initiating cells.13 The expression of b-catenin has
been shown to increase from low-grade to high-grade glio-
ma,14 and its higher expression is correlated with poorer patient
prognosis.15 17 b-catenin is known to act as a cofactor for tran-
scription factors such as T cell factor 3 (TCF3). Embryonic stem
cells, as well as poorly differentiated GBM tissues, often share a
common gene signature that includes Nanog, Oct-4, and c-Myc
expression, which is contributed by the activity of b-catenin-TCF
complex.18 Apart from the canonical Wnt-frizzled pathway,
other known upstream regulators of b-catenin activity are
EGFR16 and integrins.19 The RGD (arginine-glycine-aspartate)
motif within the carboxy terminal domain of IGFBP-2 plays a
crucial role in integrin binding and activation of integrin signal-
ing.6 Proteolytic cleavage of IGFBP-2 within its linker region re-
sults in N and C-terminal fragments, and these fragments are
known to have reduced affinity for IGFs.20 Considering this fact,
it was assumed that IGFBP-2 loses its IGF regulatory function
once cleaved. The C-terminal fragments of IGFBP-2 purified
from human hemofiltrate were found to maintain their disul-
fide bridging pattern,20 ability to bind to cell membrane, and in-
duce proliferation in rat growth plate chondrocytes.21 These
studies led us to speculate that the C-terminal domain of
IGFBP-2 may be sufficient to induce integrin-mediated func-
tions of IGFBP-2. To date, a protumorigenic function of the
C-terminal domain of IGFBP-2 has not been demonstrated.
In view of the above, the aim of this study was to delineate the
pathway involved in IGFBP-2-mediated regulation of b-catenin
and to determine if the carboxy terminal domain of IGFBP-2
has any role in b-catenin regulation and tumor growth. The cur-
rent study evaluates possible cross talk between IGFBP-2-induced
integrin signaling and b-catenin signaling in glioma, the 2 very
important pathways involved in glioma pathogenesis.
Materials and Methods
Cell Lines, Clones, and Transfection
IGFBP-21 328pcDNAmycHis, IGFBP-21 204pcDNAmycHis,
205 328
IGFBP-2 pcDNAmycHis constructs were generated by in-
serting PCR-amplified regions of full-length, N and C-terminal
domains of IGFBP-2 in pcDNAmycHis3.1(+)A. Primer sequences
are listed in Supplementary material. Table S1. U87, LN229,
U251, A172, T98G, and U373 glioma cell lines were procured
in 2013 from a European collection of cell cultures (ECACC)
and have been tested and authenticated by DNA profiling
and karyotyping. All transfections were done using Lipofect-
amine 2000 (Life Technologies, USA) as per the manufacturers
protocols. U87 and LN229 stable transfectants were estab-
lished by selection using G418 (Invitrogen).
Immunocytochemistry
Cells were allowed to grow on cover-slips, fixed with chilled
methanol for 10 minutes, incubated with primary antibody
1488
2 of 11
overnight (b-catenin, #C2206, Sigma-Aldrich, 1:50), washed
with PBS, and then incubated with 1:1500 dilution of anti-rabbit
Alexa Flour 488 secondary antibody for 1 hour; 0.5 mg/mL of
propidium iodide was used to stain nuclei.
Nuclear-cytoplasmic Fractionation, Treatments, and
Western Blotting
Downloaded from http://neuro-oncology.oxfordjournals.org/ at Mihai Eminescu Central University Library of Iasi on November 25, 2016
Cytoplasmic extract was collected after lysing cells with 0.5% of
Nonidet-40 (NP-40). The nuclear pellet was resuspended in lysis
buffer (10 mM Tris-Cl; pH 7.4, 4 mM of EDTA, 30 mM KCl, 1%
NP-40, 1 mM DTT, 100 mM of sodium ortho-vanadate, 0.5 mM
PMSF, and 1x protease inhibitor cocktail from Calbiochem). For
treatments, cells were incubated with 250 ng/mL of Wnt3a
(R&D Systems)/20 mM of LiCl/10 mM PP2 (Calbiochem)/10 mM
PP3 (Calbiochem)/10 mM RGD/10 mM focal adhesion kinase
(FAK) inhibitor (Sigma-Aldrich, USA)/10 mM IGF1R (Calbio-
chem)/500 ng/mL of IGFBP-2 (# 674-B2, R&D Systems, USA)
for 24 hours. Twenty mg of total protein was used for Western
blot analysis. The antibodies used in the study were IGF1R,
p-GSK3b, GSK3b, p-FAK, FAK, p-b-catenin, p-Akt (Ser 473), Akt
(obtained from Cell Signaling Technologies, USA); p-Akt 1/2/3
(Thr308), IGFBP-2 (Santa Cruz, USA); b-catenin and b-Actin
(Sigma-Aldrich); p-IGF1R was obtained from Abcam, UK;
lamin was purchased from IMGENEX, India. IGFBP-2 (#C-18
sc6001, Santa Cruz) and b-catenin (#C2206, Sigma-Aldrich,
USA) antibodies were used for immunohistochemistry. An
in-house generated rabbit polyclonal anti IGFBP-2 antibody
was utilized for immunohistochemistry on xenograft tissues.
Luciferase Assays
Cells were plated in 24 well dishes and after 1824 hours were
transfected with 500 ng each of IGFBP-2-pcDNAmycHis
constructs + 500 ng of Top/flash reporter construct + 10 ng of
Renilla luciferase construct (pRL-TK). Thirty-six hours after
transfection, treatments were given for 24 hours, and cells
were lysed in lysis buffer. Assays were done as per the manu-
facturers protocol (Dual-Luciferase Reporter Assay System,
Promega, USA).
Semiquantitative and Real-time RT-PCR
Two mg of RNA were reverse transcribed using a cDNA synthesis kit
(Applied Biosystems, USA). cDNA equivalent of 20 ng of RNA was
used for semiquantitative PCR. Real-time PCR reactions were per-
formed using DyNAmoColorFlash SYBR Green qPCR Kit (Thermo
Fisher Scientific India Private Ltd.) in ABI Prism 7900HT sequence
detection system (Applied Biosystems, USA). The expression of
RPL-35a gene and GAPDH (glyceraldehyde 3 phosphate dehydro-
genase) were used as normalizing controls (Primer sequences are
provided in Supplementary material. Table S2).
Development of Subcutaneous Xenografts in Nude Mice
Four million cells of respective clonal populations (LN229-Vc/
LN229-BP2/LN229-BP2-N/LN229-BP2-C) were injected into
46 week old nude mice. All animal experiments were conduct-
ed in accordance with institutional guidelines and approval of
the Institutional Animal Ethics Committee. Tumor growth was
Neuro-Oncology

monitored over a period of 7 8 weeks, after which the mice
were sacrificed and tumors dissected out, weighed, and sec-
tioned for histopathology and immunohistochemistry (IHC).
IHC and characterization are described in supplementary
information.
Patient Characteristics and Immunohistochemistry
Following institutional ethical committee approval and in-
formed patient consent, tumor tissue samples of participants
with newly diagnosed GBM (n 136) were obtained from pa-
tients who underwent surgery at the two clinical centers. The
cohort of GBM participants received standard treatment with
adjuvant radio-chemotherapy following surgery. Their survival
pattern was noted over a period of 47 months. Overall survival
was defined as the duration between surgery and the patients
death due to disease.
Tissue samples were formalin-fixed and paraffin-embedded.
Details of the IHC staining pattern for IGFBP-2 (n 136) were
retrieved from our previous study22 and were employed for fur-
ther analysis. IGFBP-2 staining was mainly cytoplasmic in the
tumor cells. Similarly, 112 of 136 cases were evaluated for
b-catenin based on tissue availability. In brief, the slides were
deparaffinized, subjected to antigen retrieval in Tris-EDTA buff-
er, incubated with primary antibody (b-catenin, #C2206,
Sigma-Aldrich, 1:100) followed by secondary antibody and vi-
sualized using betazoid diaminobenzidine (M1U539GL10,
MACH-1 polymer detection kit). A visual semiquantitative grad-
ing scale was applied to assess the intensity of immunoreactiv-
ity for GBM tissues. The staining pattern for b-catenin was both
cytoplasmic and nuclear. Only nuclear staining was considered
for evaluation. The intensity was scored on a scale of zero to
2+; where zero represented nil, 1+ represented faint, and 2+
represented strong staining. Only 2+ staining was considered
for analysis. The extent of staining in all cases was depicted
as the labeling index (LI), which was defined as the percentage
of 2+ staining tumor cells to the total number of tumor cells
counted. The median LI served as a cutoff to distinguish
cases with high expression from those with low (and absent)
expression for IGFBP-2 and b-catenin, respectively.
Statistical Analysis
For analyzing cell line data, the nonparametric unpaired t test
was utilized to calculate the P value of difference between 2 ex-
perimental datasets. All experiments with P value .05 were
considered statistically significant. SPSS 15.0 statistical software
(SPSS, Inc.) was used for clinical data analysis. Correlation of pro-
tein expression between IGFBP-2 and b-catenin was performed
using Spearmans rho test. For correlation of protein expression
with overall survival in GBMs, univariate Cox regression model
was employed. Variables that were associated with P , .10 in
univariate analyses were considered for inclusion into the multi-
variate model. Results were reported using the P values, and the
estimated hazard ratio (HR). P , .05 was considered significant,
and all exact 2-sided P values were reported. A median cutoff
was employed to distinguish tumors with high and low expres-
sion of the proteins. Further, for clinical relevance of b-catenin
and IGFBP-2, log-rank tests for significance were performed,
and Kaplan-Meier curves were generated.
Neuro-Oncology1489
Neuro-Oncology 3 of 11
Patil et al.: IGFBP-2 regulates b-catenin in GBM
Results
IGFBP-2 Regulated Stabilization of Intracellular b-catenin
To establish the role of IGFBP-2-mediated protumorigenic ac-
tions in GBM, LN229 and U87 glioma cell lines stably overex-
pressing IGFBP-2 (U87-IGFBP-2 and LN229-IGFBP-2) were
studied for b-catenin regulation. These cell lines were chosen
based on their endogenous IGFBP-2 levels.8 Western blot anal-
Downloaded from http://neuro-oncology.oxfordjournals.org/ at Mihai Eminescu Central University Library of Iasi on November 25, 2016
ysis indicated that IGFBP-2 knockdown in U251, T98G, and
U373 cell lines and overexpression in LN229 and U87 cell
lines resulted in concomitant decrease or increase in the intra-
cellular b-catenin levels respectively, within 48 hours of trans-
fection (Fig. 1A, Supplementary Material, Fig. S1A). Also, a
marked increase in intracellular b-catenin levels was observed
upon overexpression of IGFBP-2 (Supplementary Material, Figs
S1B and C). However, there was no apparent increase in the
b-catenin RNA level indicating stabilization of b-catenin protein
by IGFBP-2 (Supplementary Material, Fig. S1D). Interestingly,
there was no regulation of canonical Wnt ligands upon
IGFBP-2 knockdown/overexpression in glioma cell lines, sug-
gesting that b-catenin regulation by IGFBP-2 might be indepen-
dent of Wnt regulation (Supplementary Material, Fig. S2A).
IGFBP-2-regulated Nuclear Accumulation and Activity
of b-catenin Involves Inactivation of GSK3b
b-catenin is known for its rapid intracellular turnover by a de-
struction complex that includes axin, adenomatous polyposis
coli (APC), b-catenin, and glycogen synthase kinase (GSK3b).13
GSK3b phosphorylates b-catenin at Ser33/37/Thr41 and primes
it for proteasomal degradation.23 Phosphorylation of GSK3b at
Ser9 position renders this enzyme inactive, which subsequently
leads to stabilization of b-catenin.24 IGFBP-2 knockdown in
U251, T98G, and U373 cells resulted in decreased phosphoryla-
tion of GSK3b at Ser9 (Fig. 1A, Supplementary Material, Fig. S1A).
IGFBP-2 overexpression in LN229 and U87 cell lines consequently
resulted in increased levels of phosphorylated GSK3b (Fig. 1A).
Therefore, inactivation of GSK3b could be one of the principal
mechanisms behind IGFBP-2-mediated stabilization of b-catenin.
This IGFBP-2-mediated inactivation of GSK3b resulted in concom-
itant decrease in the levels of phosphorylated b-catenin (Fig. 1A).
AKT is known to stabilize b-catenin by inactivating GSK3b.25 How-
ever, only 2 cell lines out of 5 showed IGFBP-2-mediated regula-
tion of AKT at the Thr308 position, indicating some other
IGFBP-2-mediated upstream regulator of GSK3b (Fig. 1A, Supple-
mentary Material, Fig. S1A). Cytoplasmic stabilization of b-catenin
is often found to be associated with its increased nuclear trans-
location and transcriptional activity. IGFBP-2 overexpression in
LN229 and U87 cells resulted in increased cytoplasmic stabiliza-
tion and subsequent nuclear accumulation of b-catenin that ul-
timately reflected in the increased b-catenin-TCF reporter activity
(Fig. 1B and C).
IGFBP-2-induced b-catenin Activity Was
Integrin-signaling Dependent but IGF1R-signaling
Independent
The regulation of b-catenin by IGFBP-2 may involve IGF signal-
ing. Hence, we tested the expression of IGF-I and II upon
Patil et al.: IGFBP-2 regulates
regulates b-catenin
b-catenin in
in GBM
GBM

Downloaded from http://neuro-oncology.oxfordjournals.org/ at Mihai Eminescu Central University Library of Iasi on November 25, 2016

Fig. 1. IGFBP-2-regulated nuclear accumulation and activity of b-catenin involves inactivation of GSK3b (A) Western blots showing regulation of
GSK3b, AKT, b-catenin, and p-b-catenin upon knockdown/overexpression of IGFBP-2 within 48 hours of transfection. (B) Western blots showing
cytoplasmic as well as nuclear accumulation of b-catenin upon overexpression of IGFBP-2 in LN229 and U87 cell lines (C) Top/flash luciferase assay
showing increased b-catenin nuclear activity in IGFBP-2 overexpressing cells (n 3, P , .05). (D) Western blot showing p-IGF1R and IGF1R
expression in LN229 and U87 cell lines upon IGFBP-2 overexpression.

overexpression or knockdown in glioma cells. All glioma cell
lines under study did not show detectable mRNA levels of
IGF-I or IGF-II (Supplementary Material, Fig. S2B). In addition,
there was no regulation of p-IGF1R upon IGFBP-2 overexpres-
sion, which indicated that the regulation of b-catenin by
IGFBP-2 could be independent of IGF signaling (Fig. 1D).
IGFBP-2 is reported to activate FAK upon activation of integrin
signaling.21 Activation of a5b1 integrin signaling results in
Src-mediated phosphorylation of FAK at Tyr925 position.26,27
It is reported that a5b1 integrin signaling-mediated inactiva-
tion of GSK3b acts as a survival signal for leukemic cells.28
Moreover, integrin-activated-FAK is shown to activate
HIF-1alpha via inactivation of GSK3b in GBM.29 These earlier re-
ports suggest that activated FAK could be the upstream

1490
4 of 11 Neuro-Oncology
 Patil et al.: IGFBP-2 regulates b-catenin in GBM

regulator
monitoredofover
GSK3b. We observed
a period that IGFBP-2
of 7 8 weeks, overexpression
after which the mice Results
were obtained when these cell lines were directly treated with
in thesacrificed
were LN229 orand U87tumors
cell line resultedout,
dissected in increased levels
weighed, and of
sec- recombinant IGFBP-2 in the presence or absence of PP2 (Fig. 2B)
p-FAK,
tioned whereas inhibition ofand
for histopathology Src-mediated activation of FAK
immunohistochemistry at
(IHC). IGFBP-2 Regulated
or FAK inhibitor Stabilization
(Supplementary of Intracellular
Material, b-catenin
Fig. S3C). This indicat-
Tyr925
IHC and using Src antagonist are
characterization PP2, described
abrogated in
IGFBP-2-mediated
supplementary ed establish
To that stabilization
the role of
of b-catenin by IGFBP-2
IGFBP-2-mediated involves FAK acti-
protumorigenic ac-
inactivation
information.of GSK3b and subsequent stabilization of b-catenin vation.
tions in GBM, LN229 and U87 glioma cell linesinactivation
On the other hand, IGFBP-2-induced of
stably overex-
(Fig. 2A, Supplementary Material, Fig. S3A). PP3 treatment was GSK3b was not inhibited upon inhibition of IGF-1
pressing IGFBP-2 (U87-IGFBP-2 and LN229-IGFBP-2) were receptor sig-
used as a negative control in this experiment because PP3 does naling, which
studied for further
b-catenin confirmed
regulation. that
These cellthese effects
lines were are
chosen
Patient Characteristics
not inhibit and Immunohistochemistry
Src-mediated phosphorylation of FAK. Similar results IGF1R-signaling independent (Fig. 2B). One8 Western
hour of treatment
based on their endogenous IGFBP-2 levels. blot anal-
Following institutional ethical committee approval and in-

Downloaded from http://neuro-oncology.oxfordjournals.org/ at Mihai Eminescu Central University Library of Iasi on November 25, 2016
ysis indicated that IGFBP-2 knockdown in U251, T98G, and
formed patient consent, tumor tissue samples of participants U373 cell lines and overexpression in LN229 and U87 cell
with newly diagnosed GBM (n 136) were obtained from pa- lines resulted in concomitant decrease or increase in the intra-
tients who underwent surgery at the two clinical centers. The cellular b-catenin levels respectively, within 48 hours of trans-
cohort of GBM participants received standard treatment with fection (Fig. 1A, Supplementary Material, Fig. S1A). Also, a
adjuvant radio-chemotherapy following surgery. Their survival marked increase in intracellular b-catenin levels was observed
pattern was noted over a period of 47 months. Overall survival upon overexpression of IGFBP-2 (Supplementary Material, Figs
was defined as the duration between surgery and the patients S1B and C). However, there was no apparent increase in the
death due to disease. b-catenin RNA level indicating stabilization of b-catenin protein
Tissue samples were formalin-fixed and paraffin-embedded. by IGFBP-2 (Supplementary Material, Fig. S1D). Interestingly,
Details of the IHC staining pattern for IGFBP-2 (n 136) were there was no regulation of canonical Wnt ligands upon
retrieved from our previous study22 and were employed for fur- IGFBP-2 knockdown/overexpression in glioma cell lines, sug-
ther analysis. IGFBP-2 staining was mainly cytoplasmic in the gesting that b-catenin regulation by IGFBP-2 might be indepen-
tumor cells. Similarly, 112 of 136 cases were evaluated for dent of Wnt regulation (Supplementary Material, Fig. S2A).
b-catenin based on tissue availability. In brief, the slides were
deparaffinized, subjected to antigen retrieval in Tris-EDTA buff-
er, incubated with primary antibody (b-catenin, #C2206, IGFBP-2-regulated Nuclear Accumulation and Activity
Sigma-Aldrich, 1:100) followed by secondary antibody and vi- of b-catenin Involves Inactivation of GSK3b
sualized using betazoid diaminobenzidine (M1U539GL10,
b-catenin is known for its rapid intracellular turnover by a de-
MACH-1 polymer detection kit). A visual semiquantitative grad-
struction complex that includes axin, adenomatous polyposis
ing scale was applied to assess the intensity of immunoreactiv-
coli (APC), b-catenin, and glycogen synthase kinase (GSK3b).13
ity for GBM tissues. The staining pattern for b-catenin was both
GSK3b phosphorylates b-catenin at Ser33/37/Thr41 and primes
cytoplasmic and nuclear. Only nuclear staining was considered
it for proteasomal degradation.23 Phosphorylation of GSK3b at
for evaluation. The intensity was scored on a scale of zero to
Ser9 position renders this enzyme inactive, which subsequently
2+; where zero represented nil, 1+ represented faint, and 2+
leads to stabilization of b-catenin.24 IGFBP-2 knockdown in
represented strong staining. Only 2+ staining was considered
U251, T98G, and U373 cells resulted in decreased phosphoryla-
for analysis. The extent of staining in all cases was depicted
tion of GSK3b at Ser9 (Fig. 1A, Supplementary Material, Fig. S1A).
as the labeling index (LI), which was defined as the percentage
IGFBP-2 overexpression in LN229 and U87 cell lines consequently
of 2+ staining tumor cells to the total number of tumor cells
resulted in increased levels of phosphorylated GSK3b (Fig. 1A).
counted. The median LI served as a cutoff to distinguish
Therefore, inactivation of GSK3b could be one of the principal
cases with high expression from those with low (and absent)
mechanisms behind IGFBP-2-mediated stabilization of b-catenin.
expression for IGFBP-2 and b-catenin, respectively.
This IGFBP-2-mediated inactivation of GSK3b resulted in concom-
itant decrease in the levels of phosphorylated b-catenin (Fig. 1A).
Statistical Analysis AKT is known to stabilize b-catenin by inactivating GSK3b.25 How-
ever, only 2 cell lines out of 5 showed IGFBP-2-mediated regula-
For analyzing cell line data, the nonparametric unpaired t test
tion of AKT at the Thr308 position, indicating some other
was utilized to calculate the P value of difference between 2 ex-
IGFBP-2-mediated upstream regulator of GSK3b (Fig. 1A, Supple-
perimental datasets. All experiments with P value .05 were
mentary Material, Fig. S1A). Cytoplasmic stabilization of b-catenin
considered statistically significant. SPSS 15.0 statistical software
is often found to be associated with its increased nuclear trans-
(SPSS, Inc.) was used for clinical data analysis. Correlation of pro-
location and transcriptional activity. IGFBP-2 overexpression in
tein expression between IGFBP-2 and b-catenin was performed
LN229 and U87 cells resulted in increased cytoplasmic stabiliza-
using Spearmans rho test. For correlation of protein expression
tion and subsequent nuclear accumulation of b-catenin that ul-
with overall survival in GBMs, univariate Cox regression model
timately reflected in the increased b-catenin-TCF reporter activity
was employed. Variables that were associated with P , .10 in
(Fig. 1B and C).
univariate analyses were considered for inclusion into the multi-
variate model. Results were reported using the P values, and the
estimated hazard ratio (HR). P , .05 was considered significant, IGFBP-2-induced b-catenin Activity Was
and allIGFBP-2-induced
Fig. 2. exact 2-sided Pb-catenin
values were reported.
activity A median cutoff
was integrin-signaling dependent but IGF1R signaling-independent (A) Western blots showing
Integrin-signaling Dependent but IGF1R-signaling
was employed
regulation of FAKto
atdistinguish
Tyr925, GSK3b tumors with
at Ser9 andhigh
totaland low expres-
b-catenin in IGFBP-2 overexpressing LN229 and U87 cell lines in presence or absence of
sion PP3
of the Independent
PP2. wasproteins.
used as aFurther,
negativefor clinical
control for relevance of b-catenin
PP2. (B) Regulation of p-GSK3b when cells were treated with 500 ng/mL of IGFBP-2 in presence of
andorIGFBP-2,
PP2 log-rank
IGF1R inhibitor. (C) tests for significance
Top/flash luciferase assaywere performed,
indicating Theofregulation
the activity of b-catenin
nuclear b-catenin by IGFBP-2
in IGFBP-2 may involve
overexpressing IGF signal-
cells when treated
and Kaplan-Meier
with curves were generated.
PP2/PP3/RGD/Wnt3a/LiCl/IGF1R inhibitor (n 3, P , .05). ing. Hence, we tested the expression of IGF-I and II upon

Neuro-Oncology1491
Neuro-Oncology 5
3 of 11
Patil et al.: IGFBP-2 regulates
regulates b-catenin
b-catenin in
in GBM
GBM
with recombinant IGFBP-2 was sufficient to inactivate GSK3b,
which indicated that it could be direct regulation via a series
of phosphorylation events (Supplementary Material, Fig. S2C).
These results are further supported by a b-catenin-TCF respon-
sive luciferase reporter assay. Src inhibitor PP2 and integrin in-
hibitor RGD peptide inhibited IGFBP-2-induced luciferase
activity, while IGF1R inhibitor could not inhibit b-catenin nucle-
ar activity (Fig. 2C). Wnt3a and LiCl treatment or IGFBP-2 over-
expression resulted in increased nuclear b-catenin activity,
whereas there was no synergistic increase in the nuclear activ-
ity of b-catenin when IGFBP-2-overexpressing cells were treat-
ed with Wnt3a (Fig. 2C). This indicated that either IGFBP-2 or
Wnt3a was independently sufficient to induce the b-catenin
pathway in glioma cells and might converge at a common
b-catenin regulator, GSK3b. The effect of RGD and IGF1R inhib-
itor was confirmed by Western blot analysis for their targets
p-FAK and p-IGF1R, while the activity of Wnt3a and LiCl was
confirmed by analyzing expression levels of p-GSK3b (Supple-
mentary Material, Fig. S3B).
Carboxy Terminal Domain of IGFBP-2 Was Sufficient to
Activate b-catenin Signaling
Most of the IGF binding-independent mechanisms of IGFBP-2 are
attributed to the C-terminal RGD motif through which it is known
to activate integrin signaling.6 C-terminal fragments of IGFBP-2
purified from human hemofiltrate are known to have less IGF
binding affinity but have an intact domain structure attributed
to the disulfide bridging pattern of these fragments.20 It was pro-
posed that C-terminal fragments generated after proteolytic
cleavage of IGFBP-2 might have a biological role in cancer pro-
gression. To investigate this, we cloned and expressed N-terminal
(1 165IGFBP-2) and C-terminal domains (166 289IGFBP-2) of
IGFBP-2 separately in LN229 and U87 cell lines (Supplementary
Material, Fig. S4, S5, S6A), which have low endogenous expression
of IGFBP-2. Interestingly, stable overexpression of C-terminal
domain of IGFBP-2 was as potent as full-length IGFBP-2 in the
stabilization of b-catenin and in inducing b-catenin nuclear accu-
mulation and activity (Fig. 3AD). This was also reflected in the
regulation of b-catenin transcriptional targets Oct-4, Nanog,
MMP2, and c-Myc in these cells (Fig. 3E, Supplementary Material,
Fig. S7D). This C-domain-induced stabilization of b-catenin was
compromised when cells were treated with PP2, which suggested
that IGFBP-2 C-terminal domain alone is sufficient to induce
integrin-mediated activation of b-catenin (Supplementary Mate-
rial, Fig. S6B). Moreover the C-terminal domain of IGFBP-2 in-
duced invasion of LN229 and U87 cells (Supplementary
Material, Fig. S7AC).
C-terminal Domain of IGFBP-2 Showed Faster in vivo
Tumor Growth Than N-terminal Domain
To investigate the role of different domains of IGFBP-2 in glioma
tumor progression, LN229 cells engineered to overexpress either
full length IGFBP-2 (LN229-IGFBP-2/FL) or C (LN229-BP2-C) and
N (LN229-BP2-N) terminal domains were subcutaneously inject-
ed into immunocompromised mice. Cells transfected with clon-
ing vector (LN229-Vc/V) served as the control. As shown in Fig. 4,
subcutaneous xenografts developed from LN229 cells express-
ing full length and C-domain IGFBP-2 protein progressed faster
1492
6 of 11
than cells expressing N-terminal domain or vector control. The
tumor sizes developed from cells expressing full length (Fig. 4A
and B) or C and N terminal showed (Fig 4C E; Supplementary
Material, Fig. S8A, S8B) different sizes because these data were
obtained from different sets of experiments.
The xenograft tissue morphology was observed by hematox-
ylin and eosin staining (Fig. 4F). The immunohistochemical anal-
ysis for IGFBP-2 predominantly showed cytoplasmic staining in
each subgroup with minimal nuclear staining in 2 cases. Higher
Downloaded from http://neuro-oncology.oxfordjournals.org/ at Mihai Eminescu Central University Library of Iasi on November 25, 2016
IGFBP-2 expression was noted in LN229-BP2-C (C) and
LN229-BP2 (FL) groups as compared with LN229-BP2-N (N)
and LN229-Vc (V) groups. For b-catenin, both nuclear and cyto-
plasmic immunopositivity was noted in the samples. While the
majority of samples from the V group showed no nuclear stain-
ing, the C and FL groups displayed considerable nuclear staining
of b-catenin. Variable cytoplasmic staining was also observed in
all groups. Of all the groups studied, maximum immunopositivity
for both nuclear and cytoplasmic b-catenin staining and maxi-
mum cytoplasmic immunopositivity of IGFBP-2 was observed
in the FL group (Fig. 4F).
Coexpression of IGFBP-2 and b-catenin Correlated With
Poorer Glioblastoma Patient Prognosis
The protein expression pattern of IGFBP-2 and b-catenin was
analyzed in the tumor tissues of patients by immunohisto-
chemistry (Fig. 5A). Variable expression of IGFBP-2 for GBM
cases (as described in our previous study) was noted,22 with
the LI (Labeling Index) ranging from zero to 70% (median
LI 30%). Similarly, the nuclear expression of b-catenin was
variable, with median LI being 15%. A positive correlation be-
tween IGFBP-2 and b-catenin was noted on Spearmans rho
correlation (correlation coefficient 0.217, P .021).
The median age of the participants in this cohort was
47 years. The maximum follow-up period was 46 months. On
univariate analysis with Cox regression models, the continuous
variables such as patient age (HR: 1.022; P .014) and
b-catenin expression (HR: 1.037; P .001) were noted to influ-
ence overall survival significantly. IGFBP-2 expression demon-
strated a trend towards significance with respect to overall
survival (HR: 1.011; P .074). Subsequently, in a multivariate
Cox model, patient age (HR: 1.021; P .022) and b-catenin
(HR: 1.036; P .001) were independently associated with poor-
er prognosis in GBM patients. However, IGFBP-2 (HR: 1.006; P
.425) lost significance, indicating b-catenin to be the stronger
indicator of poor prognosis than IGFBP-2 expression.
For the purpose of clinical utility, all cases were dichoto-
mized as high (n 78) and low (n 58) expression of IGFBP-2
based on the median LI cutoff of 30% (Supplementary materi-
al, Table S3). On Kaplan-Meier survival analysis, the prognostic
association of IGFBP-2 trended towards significance (P .071;
Fig. 5B) with a median survival of 13 months in patients with
high expression of IGFBP-2 versus 18 months for those with
low expression. Similarly, with a median LI cutoff of 15% for nu-
clear b-catenin, patients with high expression of b-catenin (n
57) had a significantly worse prognosis with a median survival
of 12 months versus 19 months in patients with low expression
of b-catenin (n 55) (P .002; Fig. 5C).
Further, we evaluated the effect of coexpression of
b-catenin and IGFBP-2 and noted that a combinatorial high
Neuro-Oncology
 Patil et al.: IGFBP-2 regulates b-catenin in GBM

monitored over a period of 7 8 weeks, after which the mice Results

were sacrificed and tumors dissected out, weighed, and sec-
tioned for histopathology and immunohistochemistry (IHC).
IHC and characterization are described in supplementary
information.
Patient Characteristics and Immunohistochemistry
Following institutional ethical committee approval and in-
IGFBP-2 Regulated Stabilization of Intracellular b-catenin
To establish the role of IGFBP-2-mediated protumorigenic ac-
tions in GBM, LN229 and U87 glioma cell lines stably overex-
pressing IGFBP-2 (U87-IGFBP-2 and LN229-IGFBP-2) were
studied for b-catenin regulation. These cell lines were chosen
based on their endogenous IGFBP-2 levels.8 Western blot anal-

Downloaded from http://neuro-oncology.oxfordjournals.org/ at Mihai Eminescu Central University Library of Iasi on November 25, 2016
ysis indicated that IGFBP-2 knockdown in U251, T98G, and
formed patient consent, tumor tissue samples of participants U373 cell lines and overexpression in LN229 and U87 cell
with newly diagnosed GBM (n 136) were obtained from pa- lines resulted in concomitant decrease or increase in the intra-
tients who underwent surgery at the two clinical centers. The cellular b-catenin levels respectively, within 48 hours of trans-
cohort of GBM participants received standard treatment with fection (Fig. 1A, Supplementary Material, Fig. S1A). Also, a
adjuvant radio-chemotherapy following surgery. Their survival marked increase in intracellular b-catenin levels was observed
pattern was noted over a period of 47 months. Overall survival upon overexpression of IGFBP-2 (Supplementary Material, Figs
was defined as the duration between surgery and the patients S1B and C). However, there was no apparent increase in the
death due to disease. b-catenin RNA level indicating stabilization of b-catenin protein
Tissue samples were formalin-fixed and paraffin-embedded. by IGFBP-2 (Supplementary Material, Fig. S1D). Interestingly,
Details of the IHC staining pattern for IGFBP-2 (n 136) were there was no regulation of canonical Wnt ligands upon
retrieved from our previous study22 and were employed for fur- IGFBP-2 knockdown/overexpression in glioma cell lines, sug-
ther analysis. IGFBP-2 staining was mainly cytoplasmic in the gesting that b-catenin regulation by IGFBP-2 might be indepen-
tumor cells. Similarly, 112 of 136 cases were evaluated for dent of Wnt regulation (Supplementary Material, Fig. S2A).
b-catenin based on tissue availability. In brief, the slides were
deparaffinized, subjected to antigen retrieval in Tris-EDTA buff-
er, incubated with primary antibody (b-catenin, #C2206, IGFBP-2-regulated Nuclear Accumulation and Activity
Sigma-Aldrich, 1:100) followed by secondary antibody and vi- of b-catenin Involves Inactivation of GSK3b
sualized using betazoid diaminobenzidine (M1U539GL10,
b-catenin is known for its rapid intracellular turnover by a de-
MACH-1 polymer detection kit). A visual semiquantitative grad-
struction complex that includes axin, adenomatous polyposis
ing scale was applied to assess the intensity of immunoreactiv-
coli (APC), b-catenin, and glycogen synthase kinase (GSK3b).13
ity for GBM tissues. The staining pattern for b-catenin was both
GSK3b phosphorylates b-catenin at Ser33/37/Thr41 and primes
cytoplasmic and nuclear. Only nuclear staining was considered
it for proteasomal degradation.23 Phosphorylation of GSK3b at
for evaluation. The intensity was scored on a scale of zero to
Ser9 position renders this enzyme inactive, which subsequently
2+; where zero represented nil, 1+ represented faint, and 2+
leads to stabilization of b-catenin.24 IGFBP-2 knockdown in
represented strong staining. Only 2+ staining was considered
U251, T98G, and U373 cells resulted in decreased phosphoryla-
for analysis. The extent of staining in all cases was depicted
tion of GSK3b at Ser9 (Fig. 1A, Supplementary Material, Fig. S1A).
as the labeling index (LI), which was defined as the percentage
IGFBP-2 overexpression in LN229 and U87 cell lines consequently
of 2+ staining tumor cells to the total number of tumor cells
resulted in increased levels of phosphorylated GSK3b (Fig. 1A).
counted. The median LI served as a cutoff to distinguish
Therefore, inactivation of GSK3b could be one of the principal
cases with high expression from those with low (and absent)
mechanisms behind IGFBP-2-mediated stabilization of b-catenin.
expression for IGFBP-2 and b-catenin, respectively.
This IGFBP-2-mediated inactivation of GSK3b resulted in concom-
itant decrease in the levels of phosphorylated b-catenin (Fig. 1A).
Statistical Analysis AKT is known to stabilize b-catenin by inactivating GSK3b.25 How-
ever, only 2 cell lines out of 5 showed IGFBP-2-mediated regula-
For analyzing cell line data, the nonparametric unpaired t test
tion of AKT at the Thr308 position, indicating some other
was utilized to calculate the P value of difference between 2 ex-
IGFBP-2-mediated upstream regulator of GSK3b (Fig. 1A, Supple-
perimental datasets. All experiments with P value .05 were
Fig. 3. C-terminal domain of IGFBP-2 was sufficient to activate mentary Material,
signaling. Fig. S1A).
(A) (B) Confocal Cytoplasmic
image stabilizationofofb-catenin
indicating stabilization b-catenin in
considered statistically significant. SPSS 15.0 statistical software b-catenin
LN229
(SPSS, and
Inc.)U87
wascells stably
used overexpressing
for clinical full-length
data analysis. IGFBP-2 (FL)
Correlation of or its N or C-terminal domains (images at 63X magnification, length of thetrans-
pro-
is often found to be associated with its increased nuclear scale
bar location and transcriptional
of b-catenin inactivity. IGFBP-2 overexpression in
teinrepresents
expression20between
mm). (C) Western
IGFBP-2 blots showing nuclear
and b-catenin and cytoplasmic
was performed accumulation LN229 and U87 cells overexpressing
full-length IGFBP-2 and its N or C-terminal domains. (D) Top/flash LN229assay
luciferase and U87 cells resulted
indicating the in increased
nuclear activity cytoplasmic
of b-catenin-TCFstabiliza-
in cells
using Spearmans rho test. For correlation of protein expression
expressing full length or N or C-terminal domains of IGFBP-2 (n 3, tionP,and subsequent
.05). nuclear
(E) Real-time PCR accumulation
analysis showing of b-catenin that ul-
the regulation of
with overall survival in GBMs, univariate Cox regression model
b-catenin-TCF transcriptional targets in U87 cell line (n 3, P , .05). timately reflected in the increased b-catenin-TCF reporter activity
was employed. Variables that were associated with P , .10 in
(Fig. 1B and C).
univariate analyses were considered for inclusion into the multi-
expression
variate model.of IGFBP-2 and b-catenin
Results were was the
reported using associated
P values,with
and the survival has been depicted in Supplementary material, Table
worst prognosis
estimated hazard(median
ratio (HR).survival
P , .05of was
12 mo) as opposed
considered to tu- IGFBP-2-induced
significant, S4. These results have been validated
b-catenin Activityusing
Was the TCGA dataset,
mors
and allhaving low expression
exact 2-sided P values were or lacking
reported.both IGFBP-2cutoff
A median and where IGFBP-2 mRNA alone and coexpression of IGFBP-2 and
Integrin-signaling Dependent but IGF1R-signaling
was employed
b-catenin, which had the best
to distinguish prognosis
tumors (median
with high survival
and low expres-of b-catenin mRNA were significantly associated with poor patient
21
sionmo) (Fig.proteins.
of the 5D). Those with afor
Further, high expression
clinical of either
relevance marker Independent
of b-catenin prognosis, whereas b-catenin mRNA alone did not show a sig-
alone had an log-rank
and IGFBP-2, intermediate testsprognosis (medianwere
for significance survival of 14 The
performed, nificant association
regulation with patient
of b-catenin prognosis
by IGFBP-2 may (Supplementary Ma-
involve IGF signal-
16
andmo). The statistical
Kaplan-Meier curves weresignificance
generated.of these differences in ing. terial,Hence,
Fig. S9Aweand B, Supplementary
tested the expression material,
of IGF-ITable S5 and
and S6).
II upon

Neuro-Oncology1493
Neuro-Oncology 7
3 of 11
Patil et al.: IGFBP-2 regulates
regulates b-catenin
b-catenin in
in GBM
GBM

Downloaded from http://neuro-oncology.oxfordjournals.org/ at Mihai Eminescu Central University Library of Iasi on November 25, 2016
Fig. 4. C-terminal domain of IGFBP-2 showed faster in vivo tumor growth than N-terminal domain (A) (B) Subcutaneous injection of
LN229-IGFBP-2 (FL) cells depicting faster tumor growth in nude mice compared with LN229-Vc (Vc) cells (n 10, P , .05). (C and D)
Subcutaneous injection of LN229-BP2-C (C) cells depicting faster tumor growth in nude mice compared with LN229-BP2-N (N) or LN229-Vc (Vc)
cells (n 10, P , .05). (E) Mean xenograft tumor weight of LN229-IGFBP-2 or LN229-BP2-C compared with LN229-Vc or LN229-BP2-N (n 10, P ,
.05). (F) Hematoxylin and eosin (H and E) staining showing the tissue morphology, immunohistochemistry showing the IGFBP-2 and b-catenin
expression pattern in xenograft tumors obtained from V (original magnification x160), FL (original magnification x32, inset x160), N (160) and
C (80, inset 160) groups of nude mice. Length of the scale bar represents 20 mm.

Discussion first time, IGFBP-2 as the upstream regulator of b-catenin sig-

Despite recent advances in the treatment of GBM, the median
survival has been in the range of 13 17 months. Our previous
study,22 which corroborated other reports,3,6 suggested overex-
pression of IGFBP-2 in GBM. However, the role of IGFBP-2 in the
pathogenesis of GBM is not clear. Here we demonstrate, for the
naling in glioma. We observed that IGFBP-2 regulates stabiliza-
tion and nuclear activity of b-catenin, which was in agreement
with our earlier study in breast cancer.10 However, in contrast to
our observation in breast cancer, regulation of b-catenin was
independent of IGF1R signaling in glioma. One probable reason
for this could be that IGFBP-2 did not regulate IGFs in glioma

1494
8 of 11 Neuro-Oncology
 Patil et al.: IGFBP-2 regulates b-catenin in GBM

monitored over a period of 7 8 weeks, after which the mice Results

were sacrificed and tumors dissected out, weighed, and sec-
tioned for histopathology and immunohistochemistry (IHC).
IHC and characterization are described in supplementary
information.
Patient Characteristics and Immunohistochemistry
Following institutional ethical committee approval and in-
IGFBP-2 Regulated Stabilization of Intracellular b-catenin
To establish the role of IGFBP-2-mediated protumorigenic ac-
tions in GBM, LN229 and U87 glioma cell lines stably overex-
pressing IGFBP-2 (U87-IGFBP-2 and LN229-IGFBP-2) were
studied for b-catenin regulation. These cell lines were chosen
based on their endogenous IGFBP-2 levels.8 Western blot anal-

Downloaded from http://neuro-oncology.oxfordjournals.org/ at Mihai Eminescu Central University Library of Iasi on November 25, 2016
ysis indicated that IGFBP-2 knockdown in U251, T98G, and
formed patient consent, tumor tissue samples of participants U373 cell lines and overexpression in LN229 and U87 cell
with newly diagnosed GBM (n 136) were obtained from pa- lines resulted in concomitant decrease or increase in the intra-
tients who underwent surgery at the two clinical centers. The cellular b-catenin levels respectively, within 48 hours of trans-
cohort of GBM participants received standard treatment with fection (Fig. 1A, Supplementary Material, Fig. S1A). Also, a
adjuvant radio-chemotherapy following surgery. Their survival marked increase in intracellular b-catenin levels was observed
pattern was noted over a period of 47 months. Overall survival upon overexpression of IGFBP-2 (Supplementary Material, Figs
was defined as the duration between surgery and the patients S1B and C). However, there was no apparent increase in the
death due to disease. b-catenin RNA level indicating stabilization of b-catenin protein
Tissue samples were formalin-fixed and paraffin-embedded. by IGFBP-2 (Supplementary Material, Fig. S1D). Interestingly,
Details of the IHC staining pattern for IGFBP-2 (n 136) were there was no regulation of canonical Wnt ligands upon
retrieved from our previous study22 and were employed for fur- IGFBP-2 knockdown/overexpression in glioma cell lines, sug-
ther analysis. IGFBP-2 staining was mainly cytoplasmic in the gesting that b-catenin regulation by IGFBP-2 might be indepen-
tumor cells. Similarly, 112 of 136 cases were evaluated for dent of Wnt regulation (Supplementary Material, Fig. S2A).
b-catenin based on tissue availability. In brief, the slides were
deparaffinized, subjected to antigen retrieval in Tris-EDTA buff-
er, incubated with primary antibody (b-catenin, #C2206, IGFBP-2-regulated Nuclear Accumulation and Activity
Sigma-Aldrich, 1:100) followed by secondary antibody and vi- of b-catenin Involves Inactivation of GSK3b
sualized using betazoid diaminobenzidine (M1U539GL10,
b-catenin is known for its rapid intracellular turnover by a de-
MACH-1 polymer detection kit). A visual semiquantitative grad-
struction complex that includes axin, adenomatous polyposis
ing scale was applied to assess the intensity of immunoreactiv-
coli (APC), b-catenin, and glycogen synthase kinase (GSK3b).13
ity for GBM tissues. The staining pattern for b-catenin was both
GSK3b phosphorylates b-catenin at Ser33/37/Thr41 and primes
cytoplasmic and nuclear. Only nuclear staining was considered
it for proteasomal degradation.23 Phosphorylation of GSK3b at
for evaluation. The intensity was scored on a scale of zero to
Ser9 position renders this enzyme inactive, which subsequently
2+; where zero represented nil, 1+ represented faint, and 2+
leads to stabilization of b-catenin.24 IGFBP-2 knockdown in
represented strong staining. Only 2+ staining was considered
U251, T98G, and U373 cells resulted in decreased phosphoryla-
for analysis. The extent of staining in all cases was depicted
tion of GSK3b at Ser9 (Fig. 1A, Supplementary Material, Fig. S1A).
as the labeling index (LI), which was defined as the percentage
IGFBP-2 overexpression in LN229 and U87 cell lines consequently
of 2+ staining tumor cells to the total number of tumor cells
resulted in increased levels of phosphorylated GSK3b (Fig. 1A).
counted. The median LI served as a cutoff to distinguish
Therefore, inactivation of GSK3b could be one of the principal
cases with high expression from those with low (and absent)
mechanisms behind IGFBP-2-mediated stabilization of b-catenin.
expression for IGFBP-2 and b-catenin, respectively.
This IGFBP-2-mediated inactivation of GSK3b resulted in concom-
Fig. 5. Coexpression of IGFBP-2 and b-catenin correlated to the worst GBM itant decrease
patient in the(A)
prognosis levels of phosphorylatedshowing
Immunohistochemistry b-catenin the(Fig. 1A).
staining
25
Statistical
pattern Analysis
of IGFBP-2 and b-catenin in normal tissue and GBM patient tissues. AKTIGFBP-2
is known to stabilize
stainings are depicted by inactivating
in normal
b-catenin GSK3b.
brain tissue (a), GBM How-
with
nil expression (c), moderate expression (e), and high expression (g). Similarly, only 2 stainings
b-catenin
ever, cell linesareoutdepicted in normal
of 5 showed brain tissue (b), GBM
IGFBP-2-mediated with
regula-
Forexpression
nil analyzing(d),cell line data,
moderate the nonparametric
expression unpaired
(f), and high expression (h)t (All
testfigures areoforiginal
tion AKT x160.)
at the Higher magnification
Thr308 position,of indicating
IGFBP-2 andsome
b-catenin are
other
was utilized
shown to (g
as insets calculate the P value
and h; original of difference
magnification x320).between
Length of 2the ex-scale bar represents 20 mm. (B) Kaplan-Meier survival analysis; Prognosis of
IGFBP-2-mediated upstream regulator of GSK3b (Fig. 1A, Supple-
perimental
GBM patients datasets. All experiments
when categorized into IGFBP-2 with
low P value
and .05
IGFBP-2 highweregroups on the basis of staining intensity in IHC. (C) Prognosis of GBM patients
mentary Material, Fig. S1A). Cytoplasmic stabilization of b-catenin
considered
when statistically
categorized significant.
into b-catenin SPSS
low and 15.0 statistical
b-catenin software
high groups. (D) The IGFBP-2 low, b-catenin low category with a median survival of 21 months,
is often found to be associated with its increased nuclear trans-
(SPSS, Inc.)
whereas the was used
IGFBP-2 for b-catenin
high, clinical data
highanalysis.
categoryCorrelation
had a median of survival
pro- of only 12 months. IGFBP-2 and b-catenin coexpression in GBM tissues
location and transcriptional activity. IGFBP-2 overexpression in
tein expression
predicted patient between
prognosis IGFBP-2 and b-catenin
with statistical significancewas
of Pperformed
.001(log-rank test).
LN229 and U87 cells resulted in increased cytoplasmic stabiliza-
using Spearmans rho test. For correlation of protein expression
tion and subsequent nuclear accumulation of b-catenin that ul-
with overall survival in GBMs, univariate Cox regression model
timately reflected in the increased b-catenin-TCF reporter activity
cell
waslines, which was
employed. observed
Variables that in breast
were cancer cells.
associated withWe .10 in suggesting noncanonical regulation of b-catenin through
P ,further
delineated this pathway and observed that into
regulation (Fig. 1B and
of integrin C).
pathway. The data from this study and the previously
univariate analyses were considered for inclusion the multi-
signaling
variate model.
b-catenin Resultsby were
IGFBP-2 involves
reported integrin-mediated
using the P values, and the published report8 demonstrate a role for secreted IGFBP-2 in
inac-
tivation
estimated of hazard
GSK3b. ratio
Inactivation
(HR). P ,of.05 GSK3b is also often
was considered found to IGFBP-2-induced
significant, b-catenin regulation and protumorigenic
b-catenin Activity Was actions, respectively.
be
andassociated with misregulation
all exact 2-sided P values were of several
reported. other
A median such However, we have not examined if there is any role of intracel-
targetscutoff
Integrin-signaling Dependent but IGF1R-signaling
as
wasHIF-1alpha,
employed to Snail, Mdm2, IRS1,
distinguish tumorsMyc,with PTEN,
highand
andVDAC, which lular IGFBP-2 in this regulation, which is a remote possibility.
low expres-
may
sion ofultimately contribute
the proteins. Further, to for clinical relevance of b-catenin Independent
tumorigenesis. 30
Although our emphasis is upon the IGFBP-2-induced nuclear
andTranscriptional
IGFBP-2, log-rank upregulation
tests forofsignificance
Wnt1, Wnt2, wereor Wnt3a
performed, was The activity of b-catenin,
regulation we alsobyobserved
of b-catenin IGFBP-2amay marked accumulation
involve IGF signal-
not
and involved in thecurves
Kaplan-Meier IGFBP-2-mediated
were generated. regulation of b-catenin, of ing.b-catenin
Hence, we in the cytoplasm
tested upon overexpression
the expression of IGF-I and of II
IGFBP-2
upon

Neuro-Oncology1495
Neuro-Oncology 9
3 of 11
Patil et al.: IGFBP-2 regulates
regulates b-catenin
b-catenin in
in GBM
GBM
(Supplementary Material, Fig. S1C). There are reports that dis-
cuss the role of cytoplasmic b-catenin in stabilizing oncogenic
mRNAs such as Cox2,31 IL6, CA9, and SNAIL2.32 In concordance
with this, the protumorigenic role of these molecules has been
demonstrated in GBM.33 36
One more novel finding of this study is the demonstration of
biological activity of the C-terminal fragments of IGFBP-2 in gli-
oma tumorigenesis. It has been proposed that IGFBP-2 frag-
ments might have important biological function. 21 In the
background of upregulated IGFBP-2 and increased protease ac-
tivity in GBM tissues, it is important to study if any such mecha-
nism exists in vivo. Interestingly, C-terminal domain of IGFBP-2,
which contains the RGD motif, was sufficient to activate
b-catenin signaling and also showed faster in vivo tumor growth.
This suggests that C-terminal domain of IGFBP-2 has the protu-
morigenic role and is probably the mediator of IGFBP-2 associat-
ed GBM aggressiveness. This study advocates that under
physiological circumstances, cleaved C-terminal fragment of
Fig. 6. A schematic showing the mechanism of IGFBP-2-mediated
regulation of b-catenin pathway in glioma cells involving a5b1
integrin, FAK and GSK3b.
1496
10 of 11
IGFBP-2 might play an important role in the invasiveness of
GBM in an IGF-independent manner since IGFBP-2 is known to
lose its IGF binding affinity once cleaved. These in vitro and in
vivo experiments for the first time elucidate a mechanism of
IGFBP-2-induced b-catenin signaling and demonstrate a protu-
morigenic role for IGFBP-2 C-terminal fragment.
As mentioned before, there was a weak correlation between
IGFBP-2 and b-catenin staining on GBM tissues when the Spear-
man rank correlation coefficient was considered. One probable
Downloaded from http://neuro-oncology.oxfordjournals.org/ at Mihai Eminescu Central University Library of Iasi on November 25, 2016
reason behind this discordance could be the fact that b-catenin
stabilization and activity are regulated by multiple upstream reg-
ulators such Wnt-frizzled signaling and EGFR16 signaling. Activa-
tion of these pathways along with IGFBP-2-integrin activation
could be playing an important role in the subsequent regulation
of b-catenin signaling. Further, tissues showing the high coexpres-
sion of IGFBP-2 and b-catenin had the worst patient prognosis,
suggesting that IGFBP-2-integrin interaction might be responsible
for activation of multiple arms of signaling such as NFkB9 and
JNK5 besides b-catenin signaling, which ultimately determines
aggressiveness of tumor and the poorer prognosis of patients.
In conclusion, this study establishes IGFBP-2 as the upstream
regulator of the b-catenin pathway in GBM and demonstrates
the mechanism behind this regulation (Fig. 6). We observed
that IGFBP-2 or its C-terminal domain bind to the cell surface
integrins and activate FAK. Activation of FAK in turn leads to
the phosphorylation and inactivation of GSK3b. Inactivation of
GSK3b results in the stabilization and increased nuclear activity
of b-catenin. This IGFBP-2-induced b-catenin activity could be a
major contributor in determining IGFBP-2-mediated aggressive-
ness of GBM. Ability of C-terminal domain of IGFBP-2 to activate
b-catenin signaling and tumor growth in vivo makes it an attrac-
tive target for therapeutic intervention.
Supplementary Material
Supplementary material is available at Neuro-Oncology Journal
online (http://neuro-oncology.oxfordjournals.org/).
Acknowledgments
The results published here are in part based upon data generated by
The Cancer Genome Atlas (TCGA) pilot project established by the NCI
and NHGRI. Information about TCGA and the investigators and
institutions that constitute the TCGA research network can be found
at http://cancergenome.nih.gov/ (last accessed 30 September 2015).
We acknowledge the help of Kruthika, Priyanka Sehgal for suggestions
and Vikas for analysis of TCGA data, confocal facility and real-time PCR
facility at the Indian Institute of Science. Financial support for this
study was provided by the Department of Biotechnology, Government
of India.
SSP is a recipient of a University Grants Commission (India)
fellowship. Infrastructure support to MRDG was by the DST-FIST
program, Government of India.
Funding
Department of Biotechnology, Government of India.
Conflict of interest statement. No conflict reported by any authors.
Neuro-Oncology
 Patil et al.: IGFBP-2 regulates b-catenin in GBM

monitored over a period of 7 8 weeks, after which the mice Results

19. Burkhalter RJ, Symowicz J, Hudson LG, et al. Integrin regulation of
References
were sacrificed and tumors dissected out, weighed, and sec- beta-catenin signaling in ovarian carcinoma. J Biol Chem. 2011;
tioned
1. for histopathology
Ostrom and
QT, Bauchet L, Davis FG, immunohistochemistry (IHC).
et al. The epidemiology of glioma in IGFBP-2 Regulated23475.
286(26):23467 Stabilization of Intracellular b-catenin
IHC adults:
and characterization are described
a state of the science review. Neuro in Oncol.
supplementary
2014;16(7): 20. Mark S, Ku
To establish blerrole
the of ning
B, Ho S, et al. Diversity protumorigenic
IGFBP-2-mediated of human insulin-like
ac-
896913.
information. growth factor (IGF) binding protein-2
tions in GBM, LN229 and U87 glioma cell lines stably fragments in overex-
plasma:
2. Firth SM, Baxter RC. Cellular actions of the insulin-like growth factor primary
pressing structure,
IGFBP-2 IGF-binding properties,
(U87-IGFBP-2 and disulfide bonding
and LN229-IGFBP-2) were
binding proteins. Endocr Rev. 2002;23(6):824 854. pattern. Biochemistry. 2005;44(9):3644 3652.
Patient Characteristics and Immunohistochemistry studied for b-catenin regulation. These cell lines were chosen
3. Lin Y, Jiang T, Zhou K, et al. Plasma IGFBP-2 levels predict clinical 21.
basedKiepe D, Van
on their Der Pas A, IGFBP-2
endogenous Ciarmatori S, et8 al.
levels. Definedblot
Western carboxy-
anal-
Following institutional ethical committee
gliomas.approval and in- terminal fragments of insulin-like growth factor (IGF) binding

Downloaded from http://neuro-oncology.oxfordjournals.org/ at Mihai Eminescu Central University Library of Iasi on November 25, 2016
outcomes of patients with high-grade Neuro Oncol. 2009; ysis indicated that IGFBP-2 knockdown in U251, T98G, and
formed patient
11(5):468 consent, tumor tissue samples of participants
476. U373protein-2 exert
cell lines similar
and mitogenic activity
overexpression on cultured
in LN229 and rat
U87growth
cell
with Song
4. newly diagnosed
SW, Fuller GN, GBM
Khan (nA, et136) werean
al. IIp45, obtained from
insulin-like pa-
growth plate chondrocytes as IGF-I. Endocrinology. 2008;149(10):49014911.
lines resulted in concomitant decrease or increase in the intra-
tientsfactor
who binding
underwent surgery
protein at thebinding
2 (IGFBP-2) two clinical
protein,centers. The
antagonizes 22. Santosh
cellular V, Arivazhagan
b-catenin A, Sreekanthreddy
levels respectively, withinP,48
et al. Grade-specific
hours of trans-
cohort of GBM
IGFBP-2 participants
stimulation received
of glioma standardProc
cell invasion. treatment
Natl Acadwith
Sci expression
fection of insulin-like
(Fig. 1A, growth Material,
Supplementary factor-binding
Fig. proteins-2,
S1A). Also,23,a
USA. 2003;100(24):13970
adjuvant radio-chemotherapy 5. following surgery. Their survival and increase
marked 25 in astrocytomas: IGFBP-3
in intracellular emergeslevels
b-catenin as a strong predictor
was observed
pattern
5. was KN,
Mendes noted over
Wang GK,a Fuller
period ofet47al.months.
GN, Overall
JNK mediates survival
insulin-like uponofoverexpression
survival in patients with newly diagnosed glioblastoma.
of IGFBP-2 (Supplementary Material, Figs
was defined as the binding
growth factor durationprotein
between surgery alpha5-dependent
2/integrin and the patients Cancer Epidemiol Biomarkers Prev. 2010;19(6):13991408.
S1B and C). However, there was no apparent increase in the
glioma
death celldisease.
due to migration. Int J Oncol. 2010;37(1):143 153. 23. HagenRNA
b-catenin T, Vidal-puig A. Characterisation
level indicating of the
stabilization phosphorylation
of b-catenin of
protein
6. Tissue
Wangsamples
GK, Hu L, were formalin-fixed
Fuller GN, and paraffin-embedded.
et al. An interaction between insulin-like beta-catenin
by IGFBP-2 at the GSK-3 priming
(Supplementary site Ser45.
Material, Biochem
Fig. S1D). Biophys Res
Interestingly,
Details of the
growth IHC staining
factor-binding pattern
protein for IGFBP-2
2 (IGFBP2) (n 136)
and integrin were
alpha5 is Commun. 2002;294:324328.
22
there was no regulation of canonical Wnt ligands upon
retrieved fromfor
essential ourIGFBP2-induced
previous studycell and were Jemployed
mobility. Biol Chem.for fur-
2006; 24. Fangknockdown/overexpression
IGFBP-2 X, Yu SX, Lu Y, et al. Phosphorylation
in gliomaandcell
inactivation of
lines, sug-
ther 281(20):14085
analysis. IGFBP-214091.
staining was mainly cytoplasmic in the glycogen
gesting synthase kinase
that b-catenin 3 by by
regulation protein kinase
IGFBP-2 A. Proc
might Natl Acad
be indepen-
tumor
7. Hancells.
S, LiSimilarly,
Z, Master 112LM, of 136Exogenous
et al. cases were evaluated
IGFBP-2 for
promotes dent Sci
of USA.
Wnt 2000;97(22):11960 5.
regulation (Supplementary Material, Fig. S2A).
b-catenin based invasion,
proliferation, on tissue availability.
and In brief,
chemoresistance to the slides were
temozolomide in 25. Atkins RJ, Dimou J, Paradiso L, et al. Regulation of glycogen
glioma cellssubjected
deparaffinized, via the integrin
to antigen pathway.inBrTris-EDTA
b1-ERKretrieval J Cancer. 2014;
buff- synthase kinase-3 beta (GSK-3b) by the Akt pathway in gliomas.
111(7):14001409.
er, incubated with primary antibody (b-catenin, #C2206, IGFBP-2-regulated Nuclear Accumulation and Activity
J Clin Neurosci. 2012;19(11):15581563.
Sigma-Aldrich,
8. 1:100)
Patil SS, Railkar followed
R, Swain M, et by
al. secondary antibody
Novel anti IGFBP2 and
single vi-
chain of b-catenin
26. Involves
Focal SI, Hauck Inactivation
CR, Hsia DA, et al. v-SrcofSH3-enhanced
GSK3b interaction
variable
sualized fragment
using inhibits
betazoid glioma cell migration
diaminobenzidine and invasion.
(M1U539GL10, with Focal Adhesion Kinase at beta 1 Integrin-containing
J Neurooncol. b-catenin is known for its rapid intracellular turnover by a de-
MACH-1 polymer 2015;123(2):225235.
detection kit). A visual semiquantitative grad- Invadopodia Promotes Cell Invasion. J Biol Chem. 2002;277(15):
struction complex that includes axin, adenomatous polyposis
ing scale
9. HolmeswasKM,
applied
Annalato assess
M, Chua theCYX,
intensity
et al. of immunoreactiv-
Insulin-like growth 1248712490. 13
coli (APC), b-catenin, and glycogen synthase kinase (GSK3b).
ity forfactor-binding
GBM tissues.protein 2-driven pattern
The staining glioma progression is prevented
for b-catenin was bothby 27. Fu phosphorylates
PTCCY, Chang CTYCL, Lin S. migration and invasion byand
suppressing
blocking a clinically significant integrin,staining
integrin-linked kinase, and GSK3b b-catenin at Ser33/37/Thr41 primes
cytoplasmic and nuclear. Only nuclear was considered Src-mediated signaling pathways.
23 Mol Cell Biochem. 2014;387:
NF-kB network. it for proteasomal degradation. Phosphorylation of GSK3b at
for evaluation. TheProc Natl Acad
intensity was Sciscored
USA 2012;109(9):34753480.
on a scale of zero to 101111.
10. Sehgal zero
P, Kumar N, Praveen Kumar VR, et al. faint,
Regulation of Ser9 position renders this enzyme inactive, which subsequently
2+; where represented nil, 1+ represented and 2+ 28.
protumorigenic leadsDeto Toni-Costes F, of
stabilization Despeaux M, 24
b-catenin. Bertrand
IGFBP-2J, knockdown
et al. A Newin
represented strong pathways by insulin
staining. Only like growth
2+ staining was factor binding
considered
protein2 The
and extent
its association along U251, T98G, and U373 cells resulted insurvival
alpha5beta1 integrin-dependent pathway
decreased through
phosphoryla-
for analysis. of staining in with b-catenin
all cases in breast
was depicted GSK3beta activation in leukemic cells. PLoS One. 2010;5(3):e9807.
cancer lymph node metastasis. Mol Cancer. 2013;12(1):63. tion of GSK3b at Ser9 (Fig. 1A, Supplementary Material, Fig. S1A).
as the labeling index (LI), which was defined as the percentage 29. Skulioverexpression
N, Monferran S,inDelmas
11. IGFBP-2 LN229C, et U87
and al. Alphavbeta3/alphavbeta5
cell lines consequently
of 2+Pustaining
P, Zhang Z, Kang
tumor cellsC,toetthe
al. total
Downregulation
number ofoftumor
Wnt2 cells
and
integrins-FAK-RhoB: a novel pathway for hypoxia regulation in
beta-catenin by siRNALIsuppresses malignant glioma cell growth. resulted in increased levels of phosphorylated GSK3b (Fig. 1A).
counted. The median served as a cutoff to distinguish glioblastoma. Cancer Res. 2009;69(8):33083316.
Therefore, inactivation of GSK3b could be one of the principal
casesCancer
with Gene
high Ther. 2009;16(4):351
expression 361.with low (and absent)
from those
30. Sutherland
mechanisms C. What
behind Are the bona fide
IGFBP-2-mediated GSK3 Substrates?
stabilization Int J
of b-catenin.
12. Roth W,for
expression Wild-Bode
IGFBP-2 C,and
Platten M, et al.respectively.
b-catenin, Secreted Frizzled-related
Alzheimers Dis. 2011;2011:505607.
This IGFBP-2-mediated inactivation of GSK3b resulted in concom-
proteins inhibit motility and promote growth of human
malignant glioma cells. Oncogene. 2000;19(37):42104220. itant Kim
31. decrease
I, KwakinH,the
Lee levels ofb-Catenin
HK, et al. phosphorylated
recognizes (Fig.motif
a specific RNA
b-catenin 1A).
25
Statistical
13. Nager M,Analysis
Bhardwaj D, Cant C, et al. b-Catenin Signalling in AKT isinknown to stabilize b-catenin
the cyclooxygenase-2 mRNA 3by
inactivating
-UTR GSK3b.
and interacts with How-
HuR in
Glioblastoma ever, colon
only 2 cancer cells. out
cell lines Nucleic
of 5Acids Res. 2012;40(14):68636872.
showed IGFBP-2-mediated regula-
For analyzing cell Multiforme
line data, and
the Glioma-Initiating
nonparametric Cells. Chemother
unpaired t test
Res Pract. tion DUva
32. of AKTG,atBertoni
the Thr308 position,
S, Lauriola M, etindicating some other
al. Beta-catenin/HuR
was utilized to2012;2012:192362.
calculate the P value of difference between 2 ex-
14. Schule R,datasets.
Dictus C, Campos B, et al. Potential IGFBP-2-mediated upstream
post-transcriptional regulatorgoverns
machinery of GSK3b (Fig. 1A,
cancer Supple-
stem cell
perimental All experiments withcanonical
P value wnt pathway
.05 were
activation in high-grade astrocytomas. ScientificWorldJournal. 2012; mentary Material,
features Fig. S1A).
in response Cytoplasmic
to hypoxia. stabilization
PLoS One. of b-catenin
2013;8(11):e80742.
considered statistically significant. SPSS 15.0 statistical software
is often
33. foundY,toLiubeG,associated
Akasaki withetits
Chung NHC, al.increased
Inductionnuclear trans-T
of a CD4+
(SPSS,2012:697313.
Inc.) was used for clinical data analysis. Correlation of pro-
15. Rossi M, Magnoni L, Miracco C, et al. and Gli1 location and transcriptional
regulatory type 1 responseactivity. IGFBP-2 overexpression in
by cyclooxygenase-2-overexpressing
tein expression between IGFBP-2 andb-catenin
b-catenin wasareperformed
prognostic
LN229 and U87
glioma. cells resulted
J Immunol. in increased
2004;173(7):4352 cytoplasmic stabiliza-
4359.
usingmarkers in glioblastoma.
Spearmans rho test.Cancer Biol Ther. 2011;11(8):753761.
For correlation of protein expression
tion and
34. subsequent
McFarland BC, Hongnuclear accumulation
SW, Rajbhandari R, et of
al. b-catenin thatIL-6
NF-kB-induced ul-
with Yang
16. overall
W,survival
Xia Y, JiinH,GBMs,
et al. univariate Coxregulates
Nuclear PKM2 regression model
b-catenin
transactivation upon EGFR timately reflected
ensures STAT3in the increasedand
activation b-catenin-TCF reporter activity
tumor aggressiveness in
was employed. Variables thatactivation. Nature. 2011;480(7375):
were associated with P , .10 in
118122. (Fig. 1B and C). PLoS One. 2013;8(11):e78728.
glioblastoma.
univariate analyses were considered for inclusion into the multi-
variate
17. model.
Kahlert UD,Results
Maciaczykwere reported S,using
D, Doostkam et al.the P values,
Activation and the
of canonical 35. Said HM, Hagemann C, Staab A, et al. Expression patterns of the
WNT/b-catenin
estimated signaling
hazard ratio (HR).enhances
P , .05 inwas
vitroconsidered
motility of glioblastoma
significant, hypoxia-related genes osteopontin, CA9, erythropoietin, VEGF
cells by activation
IGFBP-2-induced
and HIF-1alpha b-catenin Activity
in human glioma Wasand in vivo. Radiother
in vitro
and all exact 2-sided Pofvalues
ZEB1 and
wereother activators
reported. of epithelial-
A median cutoff
to-mesenchymal transition. Cancer Lett. 2012;325(1):4253. Integrin-signaling Dependent
Oncol. 2007;83(3):398 405. but IGF1R-signaling
was employed to distinguish tumors with high and low expres-
18. Ben-Porath
sion of I, Thomson
the proteins. MW,
Further, forCarey VJ, et
clinical al. An embryonic
relevance stem
of b-catenin Independent
36. Kubelt C, Hattermann K, Sebens S, et al. Epithelial-to-
cell-like gene
and IGFBP-2, expression
log-rank signature
tests for in poorly
significance weredifferentiated
performed, mesenchymal
The regulation transition inbypaired
of b-catenin human
IGFBP-2 mayprimary
involveand
IGFrecurrent
signal-
aggressive human
and Kaplan-Meier tumors.
curves wereNat Genet. 2008;40(5):499 507.
generated. glioblastomas. Int J Oncol. 2015;46(6):2515 2525.
ing. Hence, we tested the expression of IGF-I and II upon

Neuro-Oncology1497
Neuro-Oncology 11
3 of 11