GDNF mediates glioblastoma-induced microglia attraction but not astrogliosis

14
1 23 Acta Neuropathologica Pathology and Mechanisms of Neurological Disease ISSN 0001-6322 Acta Neuropathol DOI 10.1007/s00401-013-1079-8 GDNF mediates glioblastoma-induced microglia attraction but not astrogliosis Min-Chi Ku, Susanne A. Wolf, Dorota Respondek, Vitali Matyash, Andreas Pohlmann, Sonia Waiczies, Helmar Waiczies, Thoralf Niendorf, et al.

Transcript of GDNF mediates glioblastoma-induced microglia attraction but not astrogliosis

1 23

Acta NeuropathologicaPathology and Mechanisms ofNeurological Disease ISSN 0001-6322 Acta NeuropatholDOI 10.1007/s00401-013-1079-8

GDNF mediates glioblastoma-inducedmicroglia attraction but not astrogliosis

Min-Chi Ku, Susanne A. Wolf, DorotaRespondek, Vitali Matyash, AndreasPohlmann, Sonia Waiczies, HelmarWaiczies, Thoralf Niendorf, et al.

1 23

Your article is protected by copyright and

all rights are held exclusively by Springer-

Verlag Berlin Heidelberg. This e-offprint is

for personal use only and shall not be self-

archived in electronic repositories. If you

wish to self-archive your work, please use the

accepted author’s version for posting to your

own website or your institution’s repository.

You may further deposit the accepted author’s

version on a funder’s repository at a funder’s

request, provided it is not made publicly

available until 12 months after publication.

ORIGINAL PAPER

GDNF mediates glioblastoma-induced microglia attractionbut not astrogliosis

Min-Chi Ku • Susanne A. Wolf • Dorota Respondek • Vitali Matyash •

Andreas Pohlmann • Sonia Waiczies • Helmar Waiczies • Thoralf Niendorf •

Michael Synowitz • Rainer Glass • Helmut Kettenmann

Received: 4 October 2012 / Accepted: 9 January 2013

� Springer-Verlag Berlin Heidelberg 2013

Abstract High-grade gliomas are the most common pri-

mary brain tumors. Their malignancy is promoted by the

complex crosstalk between different cell types in the cen-

tral nervous system. Microglia/brain macrophages infiltrate

high-grade gliomas and contribute to their progression. To

identify factors that mediate the attraction of microglia/

macrophages to malignant brain tumors, we established a

glioma cell encapsulation model that was applied in vivo.

Mouse GL261 glioma cell line and human high-grade

glioma cells were seeded into hollow fibers (HF) that allow

the passage of soluble molecules but not cells. The glioma

cell containing HF were implanted into one brain hemi-

sphere and simultaneously HF with non-transformed

fibroblasts (controls) were introduced into the contralateral

hemisphere. Implanted mouse and human glioma- but not

fibroblast-containing HF attracted microglia and up-regu-

lated immunoreactivity for GFAP, which is a marker of

astrogliosis. In this study, we identified GDNF as an

important factor for microglial attraction: (1) GL261 and

human glioma cells secret GDNF, (2) reduced GDNF

production by siRNA in GL261 in mouse glioma cells

diminished attraction of microglia, (3) over-expression of

GDNF in fibroblasts promoted microglia attraction in our

HF assay. In vitro migration assays also showed that

GDNF is a strong chemoattractant for microglia. While

GDNF release from human or mouse glioma had a pro-

found effect on microglial attraction, the glioma-induced

astrogliosis was not affected. Finally, we could show that

injection of GL261 mouse glioma cells with GDNF

knockdown by shRNA into mouse brains resulted in

reduced tumor expansion and improved survival as com-

pared to injection of control cells.

Keywords Glioblastoma � Microglia � GDNF �Astrocyte

Introduction

High-grade gliomas (HG-gliomas) are most aggressive

primary tumors of the central nervous system (CNS). HG-

gliomas are not a homogenous population of tumor cells,

but contain microglia which significantly contribute to the

actual tumor mass and are surrounded by activated astro-

cytes [4]. Microglia density in HG-gliomas can amount up

to 30 % of the tumor mass [25, 34]. However, unlike in an

inflamed tissue, immune functions in tumor-associated

microglia are often suppressed [10, 13, 32]. Therefore,

microglia do not attack glioma cells, but rather have been

shown to promote glioma growth. Altogether, the recent

Electronic supplementary material The online version of thisarticle (doi:10.1007/s00401-013-1079-8) contains supplementarymaterial, which is available to authorized users.

M.-C. Ku � S. A. Wolf � D. Respondek � V. Matyash �H. Kettenmann (&)

Department of Cellular Neuroscience, Max Delbruck Center

for Molecular Medicine (MDC), Robert Rossle Str. 10,

13125 Berlin, Germany

e-mail: [email protected]

A. Pohlmann � S. Waiczies � H. Waiczies � T. Niendorf

Berlin Ultrahigh Field Facility (B.U.F.F.), Max Delbruck Center

for Molecular Medicine (MDC), 13125 Berlin, Germany

M. Synowitz

Department of Neurosurgery, Charite-Universitatsmedizin

Berlin, 13353 Berlin, Germany

R. Glass

Klinikum der Universitat Munchen (LMU), Klinik fur

Neurochirurgie, Neurochirurgische Forschung, Marchioninistr.

15, 81377 Munich, Germany

123

Acta Neuropathol

DOI 10.1007/s00401-013-1079-8

Author's personal copy

data by others and us indicate that blockade of microglia

infiltration into HG-gliomas could be a novel therapeutic

target in neuro-oncology [25, 26]. Earlier studies have

shown that reactive astrocytes are found in the glioma

environment and are associated with increased glioblas-

toma invasion [8]. Moreover, reactive astrocytes are highly

concentrated at the perivascular region of HG-gliomas and

may also contribute to glioma growth, e.g. mediated by the

release of TGF-b [6].

Recent studies focus on the factors that regulate the

cross talk between glioma cells and glial cells. There is a

positive correlation between the number of microglia and

the expression level of monocyte chemoattractant protein-3

(MCP-3) [28] or granulocyte/macrophage colony-stimu-

lating factor (GM-CSF) [18]. Hepatocyte growth factor [2]

and MCP-1 had also been identified as chemoattractant in

the glioma context [21, 29].

Glial-derived neurotrophic factor (GDNF) is a potent

trophic factor for a variety of neuronal cell populations in

the CNS. GDNF was first discovered in the conditioned

medium of the B49 rat glioma cell line [22]. Importantly,

GDNF is highly expressed in human HG-gliomas [35]. It is

not only a trophic factor, but also a chemoattractant for

various types of cells [5, 7, 17, 23, 33]. Although GDNF

has been studied for several years, and the correlation of

GDNF and glioma has been addressed, the role of its potent

chemoattractant factor in the crosstalk between glioma and

microglia has not been investigated.

In this study, we investigated the role of GDNF in

facilitating the attraction of microglia to the glioma. To

distinguish between the impact of soluble factors and cell

to cell contact, we have developed a method of implanting

encapsulated glioma cells into the mouse brains. Semi-

permeable hollow fibers (HF) allow the bi-directional

exchange of oxygen, nutrients, and growth factors, but

prevent physical contact between the encapsulated cells

and their environment [3, 37]. We show here that encap-

sulated glioma cells attract microglia and induce

astrogliosis. While GDNF does not regulate astrogliosis, it

strongly mediates microglia attraction to HG-gliomas and

promotes tumor growth.

Materials and methods

Cell culture

GL261 murine glioma cells (National Cancer Institute,

Bethesda, USA) and murine fibroblast cell (SCRC-1008,

ATCC) were cultured in DMEM with 200 mM glutamine,

50 units/ml penicillin, 50 lg/ml streptomycin, and 10 %

FBS (all from Invitrogen, Darmstadt, Germany). The

human glioblastoma cells (Tumor 1 and 2) were derived

from human tumor resections and cultured in RPMI 1640

with 20 lg/ml EGF, FGF (Cell Systems, Troisdorf, Ger-

many) and supplements. Microglia was prepared from

neonatal C57BL/6 mice as previously described [25]. All

cells were maintained in a 37 �C incubator with a 5 % CO2

humidified atmosphere.

Cell encapsulation

SCRC-1008, GL261, and human glioma cells (Tumor 1)

were infused into HF (Minntech, Minneapolis, USA) at a

density of 2 9 106 cells/ml. HFs were sealed at the ends in

5-cm segments. 1 day before measurements or transplan-

tation, the 5-cm fibers were subsequently cut into 5 mm

length and sealed at both ends. Those HFs were further

sectioned in a cryostat with 10 lm thickness. Hematoxylin

and Eosin (H&E) staining’s of fiber sections were

performed.

Cell proliferation

Cell proliferation and viability were monitored using the

Alamar blue assay following manufacturer’s recommended

procedures (Invitrogen). The SCRC-1008, GL261, and

human glioma cells (Tumor 1) were seeded in 96 well

plates with 5,000 cells/100 ll density. At day 1, 3, 5, and 7

of culture, 10 ll of Alamar blue solution was added to each

well to make a 1:10 dilution and incubated for 4 h at 37 �C

in a humidified atmosphere of 5 % CO2 and 95 % air.

Before transplanting the fibers into the mouse brains, HF

with SCRC-1008, GL261, and human glioma cells (Tumor

1) were cut into 5 mm of length and sealed at both ends.

The relative cell numbers in the fiber was also measured

and only fibers with the equal amount of cells were used in

a given experiment. Wells containing empty fibers were

used as control. Absorbance of Alamar blue was measured

in a spectrometer (M200, Tecan, Crailsheim, Germany) at

570 and 600 nm.

HF inoculation

Female C57BL/6 (8–10 weeks) mice were handled

according to governmental (LAGeSo) and internal (MDC)

rules and regulations. Briefly, after anesthetized, the mouse

head was placed onto a stereotactic frame (David Kopf

Instruments, Tujunga, USA). Through a midline incision, a

burr hole was made by drilling at 1 mm anterior to the

bregma and 1.5 mm both right and left side from the

midline. Canals were created by inserting a 26-gauge

Hamilton syringe 5 mm ventral from dura mater. Then the

HFs were inserted into both canals.

Acta Neuropathol

123

Author's personal copy

Immunofluorescent staining and image processing

Mouse brains were prepared as previously described [26].

Human glioblastoma biopsy specimens were obtained from

glioblastoma patients, according to governmental and

internal (Charite) rules and regulations. Cortical brain tissue

from epilepsy patients was performed according to the rules

laid down by the Ethical Committee (Charite, EA1/142/05),

and informed consent was obtained according to the Dec-

laration of Helsinki (BMJ, 1991; 302: 1194). Primary

antibodies were added overnight at a dilution of 1:750 for

Iba-1 (WAKO, Tokyo, Japan), 1:1,000 for GFAP (Sigma-

Aldrich, Munich, Germany), 1:100 for Vimentin (Abcam,

Cambridge, UK) and 1:100 for GDNF (Santa Cruz, Hei-

delberg, Germany) at 4 �C. Alexa 488, 594-conjugated goat

anti-rabbit IgG (1:200, Invitrogen) and Cy5-conjugated

donkey anti-rabbit IgG (1:200, Jackson Lab, Hamburg,

Germany) were subsequently applied. The nuclei were

counterstained with 40,6-diamidino-2-phenylindole (DAPI,

Sigma-Aldrich).

Images were taken using a confocal microscope (TCS

SP5, Leica, Wetzlar, Germany) with 209 or 409 oil

objectives. Iba-1 positive labeled cells were counted using

ImageJ software (NIH, Bethesda, USA). The intensity of

GFAP fluorescence was measured by ImageJ without

adjusting digital gain.

GDNF knockdown and over-expression

GDNF and scrambled non-targeted or GAPDH small inter-

fering RNA (siRNA) were transfected with Dharmafect 4

according to manufacturer’s instructions (Dharmacon,

Lafayette, USA). After 24 h, GDNF or non-targeted siRNA

treated GL261 cells were injected into HF and an aliquot was

used for quality control by RT-PCR and ELISA.

GDNF transient transfection of SCRC-1008 was per-

formed using the lipofectamine LTX (Invitrogen). Plasmid

cDNA encoding GDNF was kindly provided by Dr. Carlos F.

Ibanez at the Karolinska Institute, Sweden. PCDNA3 plasmid

DNA was used as negative control.

For stable GDNF knockdown, GDNF OmicsLinkTM short

hairpin RNA (shRNA) clone (GCCAGTGTTTATCTGAT

AC) and non-targeting scrambled shRNA (shNT) were

transfected into GL261 cells according to manufacturer’s

instructions (GeneCopoeia, Rockville, USA). Transfected

GL261 cells were selected by treating with 5 lg/ml puromy-

cin. After selection, GDNF secretion was measured by ELISA.

RT-PCR and ELISA

Sequences for PCR primer sets were as follow: for GDNF

forward 50-TATGGGATGTCGTGGCTGT-30, reverse 50-CGTCATCAAACTGGTCAGGA-30, GFRa-1 forward 50-TT

CCCACACACGTTTTACCAC-30, reverse 50-TTTGTGG

TTATGTGGCTGGAG-30, GFRa-2 forward 50-AAGGC

CAACAACTCCAAAGAG-30, reverse 50-CCAAGGTCAC

CATCAGGAGT-30, RET forward 50-CCTCCGTGACAG

CCGCAAGA-30, reverse 50-CGGAATCCGGCCCTTGCT

TT-30, and for actin forward 50-CCCTGAAGTACCCCAT

TGAA-30, reverse 50-GTGGACAGTGAGGCCAAGAT-30.Supernatant collected from SCRC-1008, GL261, mouse

primary microglia, astrocytes, NPCs and human glioma

cells (Tumor 1 and 2) were determined for GDNF by

ELISA kits according to the manufacturer’s protocols

(Promega, Madison, USA). Results are presented as pico-

grams of GDNF per 50,000 cells.

Chemotaxis assays

The agarose spot assay was described before in [36]. Two

spots containing GDNF and two containing PBS were

placed onto 35-mm glass dishes (MatTek, Ashland, USA).

After 3 and 7 h, microglial cells under the spot were

counted. For time-lapse microscopy, the dish was placed

into the BioStation IM (Nikon, Dusseldorf, Germany) and

the resulting stacks were analyzed by ImageJ.

The chemotaxis of microglia was also determined using

a transwell assay with an 8-lm pore size membrane (BD

Biosciences, Heidelberg, Germany). The upper chamber of

a transwell insert was filled with 2 9 105 microglia in

suspension, and the lower chamber (BD Biosciences) with

0, 500, or 1,000 ng/ml recombinant GDNF (Peprotech,

Hamburg, Germany) or GL261 conditioned medium. After

24 h, microglial migration was quantified by counting the

number of cells that migrated through the membrane using

an inverted bright-field microscope. Five fields were quanti-

fied for each well.

In vivo assessment of glioma growth by magnetic

resonance imaging (MRI)

A total number of 10-mice were imaged at day 14 after the

bilateral intracereberal inoculation of either control (shNT)

or shGDNF GL261 cells. MR imaging was performed on a

9.4 Tesla small animal MRI system (Biospec 94/20, Bruker

BioSpin, Ettlingen, Germany) equipped with CryoProbe

(400 MHz, Bruker BioSpin). T2-weighted images (RARE,

effective echo time = 60 ms; repetition time = 3,268 ms;

RARE factor = 12) were acquired with the same slice

geometry (FOV = 18 9 18 mm, matrix size = 350 9

350, slice thickness = 270 lm, in-plane spatial resolu-

tion = 51 lm, 21 coronal slices covering a brain region of

5.67 mm starting at the frontal end of the cerebral cortex

(approx. Bregma 3.56 to -2.11 mm) [9]. Tumor volume

was calculated in mipav (http://mipav.cit.nih.gov) by add-

ing up the voxel volumes.

Acta Neuropathol

123

Author's personal copy

Survival studies were performed as previously described

[11]. Two groups of wild-type C57BL/6 mice received

either shNT or shGDNF GL261 cells. Survival rate was

calculated by the Kaplan–Meier method (MedCalc, Mari-

akerke, Belgium) with log rank analysis.

Statistical analysis

All data represent the average of at least triplicate samples.

Error bars represent standard error of the mean. Data were

analyzed by Student’s t test and one way ANOVA in

Microsoft Office Excel 2007 and applied Bonferroni cor-

rection. The differences were considered statistically

significant at p \ 0.05.

Results

Human and mouse glioma cells survive and grow in HF

To explore if soluble factors released from glioma cells

have an impact on intrinsic brain cells, we encapsulated

glioma cells in semi permeable hollow fibers as an exper-

imental model (Fig. 1b). First, we analyzed the

proliferation rate of fibroblasts (SCRC), mouse glioma

cells (GL261), and human glioma cells (Tumor 1) in 96

well plates using the Alamar blue assay. When analyzed at

day 1, 3, 5 and 7, it is apparent that all cells proliferated

with GL261 at the highest rate (Fig. 1a). Next, we tested

whether cells can survive and even proliferate in HF. HF

enclosed GL261, primary human glioma cells, and fibro-

blast were tested in different cell densities (ranging from

1 9 105 to 4 9 106 cells/ml). The optimized cell density

was 2 9 106/ml for all tested cell preparations. H&E

staining of fiber sections demonstrated the distribution and

morphology of cells in the HF (Fig. 1c). At 1, 3, 5, and

7 days after encapsulation, HF with GL261 glioma cells,

human glioma cells, or SCRC-1008 fibroblast cells showed

an increase in absorbance indicative of an increase in cell

number similar as in the 96 well plates (Fig. 1c).

Encapsulated glioma cells attract microglial cells

and trigger astrogliosis

To address the question whether soluble factors released

from encapsulated glioma cells induce microgliosis or as-

trogliosis, HF filled with GL261 glioma cells or non-

tumorigenic SCRC-1008 fibroblast cells were implanted

into mouse brains. After 2 weeks, brains were fixed and the

histological analysis of the implants showed that the outer

membrane of the mouse glioma (GL261) HF, was com-

pletely surrounded by Iba-1? microglia at a higher density

than the fibroblasts (control) HF (Fig. 2a). The density of

microglia surrounding the mouse glioma HF was 1.7-fold

higher (p \ 0.01) as compared to the fibroblast HF

(Fig. 2b). There was also a morphological transition of

microglia towards an ameboid phenotype, which was more

pronounced in cells closer to the GL261 HF (Fig. 2a, inset).

We also encapsulated HG-human glioma cells in HF and

implanted them into mouse brains. Similar to the encap-

sulated mouse glioma cells, Iba-1? microglia were found to

accumulate around the HF, indicating microgliosis

(Fig. 2a, right panel). The density of microglia surrounding

the HF containing human glioma cells was 2.5-fold higher

(p \ 0.001) as compared to the control HF (Fig. 2b).

To investigate whether soluble factors from glioma cells

also induce astrogliosis, we measured GFAP immunoflu-

orescence intensity. The density of GFAP-positive

astrocytes surrounding GL261 HF was 1.3-fold (p \ 0.05)

higher when compared with control HF (Fig. 2c, d). The

morphology of GFAP labeled astrocytes close to the mouse

glioma HF was characterized by a hypertrophic phenotype

(increase in the thickness of processes and enlarged soma)

Fig. 1 Characterization of hollow fiber cell encapsulation. a Cell

proliferation in fibroblasts, GL261 cells and human glioma cells

cultured in 96-well plates, was determined by the Alamar blue assay

1, 3, 5, and 7 days after plating. b Schematic representation of cell

encapsulation in a hollow fiber in a cross sectional view. c Histological

longitudinal cross section of hollow fibers containing encapsulated

mouse fibroblast cells (SCRC-1008), mouse GL261 glioma cells and

human glioma cells (Tumor 1). Fiber sections were visualized by

H&E staining on the 7th day of culture. Cell proliferation in HF on 1,

3, 5, and 7 days was determined by the Alamar blue assay. Bars show

mean ± SEM from three experiments

Acta Neuropathol

123

Author's personal copy

that is typical of reactive astrocytes. Astrocytes in the

vicinity of control HF had a normal morphological

appearance. Human glioma HF in the mouse brains also

resulted in astrogliosis (Fig. 2c, d). These findings suggest

that glioma cells also influence astrocytes, although not as

potently as microglia, via soluble factors.

GDNF is highly expressed in glioma

To confirm that GDNF is produced and released from the

mouse and human glioma cell lines used in our study, we

further determined GDNF mRNA and protein levels. RT-

PCR shows that GL261 cells expressed much higher

mRNA levels than mouse astrocytes or microglia while we

detected no expression in neural precursor cells (NPCs) or

fibroblasts (Fig. 3a). We detected mRNA for GDNF

expression in human glioma cells from two different

patients (Tumor 1 and 2) (Fig. 3a). GDNF secretion was

then determined by ELISA and indicated that both HG-

human glioma cells (Tumor 1 and 2) and mouse glioma

cells (GL261) secret a higher amount of GDNF. GDNF

secretion levels were significantly lower in microglia and

astrocytes, and for SCRC-1008 fibroblasts even below the

level of detection (Tumor 1, 39.3 ± 4.9; Tumor 2,

47.9 ± 4.9; GL261, 42.3 ± 6.8; astrocyte, 13.7 ± 3.5;

microglia 7.6 ± 4.6 pg/50,000 cells/24 h) (Fig. 3b).

To determine if GDNF is expressed in the tumor envi-

ronment, we implanted mCherry expressing GL261 cells

into mouse brains. As shown in Fig. 3c, GDNF is highly

expressed by GL261 cells while we detected no expression

in non-tumor area. Human glioma biopsies also showed

high expression of GDNF while tissue from human cortex

(from an epilepsy surgery) did not express GDNF. These

results show a differential expression of GDNF between

glioma cells and the glioma-associated brain cells and

suggest GDNF as a candidate molecule mediating glioma-

glial cell interaction.

GDNF plays a key role in regulating microglia

attraction

To address whether GDNF mediates glioma-glial cell

interactions, we first analyzed GDNF receptor expression

on microglia. Figure 3a shows that microglia expressed

both GDNF receptors, GFRa-1 and GFRa-2, but not the

co-receptor RET. Astrocytes, NPCs, and fibroblasts

expressed GFRa-1 and GFRa-2, and glioma cells pre-

dominantly expressed GFRa-1 (Fig. 3a).

We next impaired GDNF expression in glioma cells by

an siRNA approach. GDNF gene expression in GL261 was

reduced by [80 % after 2 days (data not shown). GDNF

secretion was examined after siRNA treatment. Figure 4a

Fig. 2 Human and mouse glioma cells encapsulated in hollow fibers

induce microglia attraction and astrogliosis. a Fibroblast cells (SCRC-

1008), mouse glioma cells (GL261), or human glioma cells were

filled in HF and transplanted into the left or right hemisphere,

respectively. After 14 days, mice were killed and brain sections were

stained by immunofluorescence with Iba-1 (green). Cell nuclei were

counterstained with DAPI (blue). The yellow dashed line indicates

the border between HF and brain tissue. b Quantitative analysis of

Iba-1? cells surrounding GL261 HF (mean ± SEM, n = 9 mice;

**p \ 0.01) and human glioma HF (mean ± SEM, n = 13 mice;

***p \ 0.001) compared with fibroblast (control) HF. c Mouse brain

sections were stained with GFAP (red). d GFAP fluorescence

intensity measured from three random areas surrounding the GL261

HF (mean ± SEM, n = 8 mice; *p \ 0.05 compared with control)

and human glioma cells (mean ± SEM, n = 13 mice; *p \ 0.05

compared with control HF)

Acta Neuropathol

123

Author's personal copy

shows that the GDNF secretion was reduced by 66 %

(p \ 0.05) in GL261 cells (siGDNF) after 6 days of GNDF

siRNA treatment while non-targeted siRNA did not affect the

secretion of GDNF (Fig. 4a). On the next day of GDNF-

silencing, GL261 glioma cells were encapsulated into HF.

The depletion of GDNF by siRNA did not affect the glioma

cell proliferation in HF (supplemental Fig. 1). We then

implanted control (siNT) and siGDNF expressing GL261 cell

containing HF into mouse brains. 6 days after implantation,

we found that GDNF knockdown reduced microglia attrac-

tion (643 ± 47 cells/mm2) to 64 % (p \ 0.001) when

compared to control (siNT, 998 ± 48 cells/mm2) (Fig. 4b, c).

As a second approach, we used stable GDNF silencing by

shRNA approach to confirm the siRNA silencing results.

Either non-targeted shRNA (shNT) or GDNF shRNA

(shGDNF) treated GL261 cells were filled into HF. 6 days

after implantation, similar results show that GDNF knock-

down reduced microglia attraction (shGDNF, 734 ± 36

cells/mm2) to 65 % (p \ 0.01) when compared to control

(shNT, 1,124 ± 79 cells/mm2) (supplemental Fig. 2).

In addition, we have tested two lines of HG-human

glioma cells which secrete different levels of GDNF

(Fig. 3a, b). Tumor 1 and Tumor 2 HG-human glioma cells

were encapsulated in HF and implanted into mouse brains.

Similar to the encapsulated mouse glioma cells, Iba-1+

microglia were found to accumulate around both HF. The

density of microglia surrounding the Tumor 1 and Tumor 2

HF was not significantly different (data not shown).

To test whether over-expression of GDNF by non-gli-

oma cells could attract more microglial cells, SCRC-1008

fibroblast cells were transfected with either empty vector

(?vector) or GDNF vector (?GDNF). GDNF mRNA

expression in SCRC-1008 was increased by [80 % after

3 days of transfection (data not shown). GDNF secretion

was significantly increased in ?GDNF cells as compared to

control (?vector, 296.7 ± 16.8 pg/50,000 cells, p \ 0.01)

(Fig. 4d). Both ?vector and ?GDNF SCRC cells were

encapsulated into HF, implanted into mouse brains and

analyzed after 6 days. Iba-1 staining revealed that the mi-

croglial density increased by 1.7-fold (p \ 0.01) around

?GDNF HF, as compared to control (Fig. 4e, f). These

results showed that GDNF released from GL261 glioma

cells and GDNF overexpressing non-tumorigenic fibro-

blasts attract microglial cells and indicate the microglia

attraction is partly GDNF dependent.

GDNF does not affect glioma-induced astrogliosis

We next evaluated the impact of GDNF on astrogliosis.

Control (siNT) versus GDNF knockdown (siGDNF)

Fig. 3 GDNF expression in glioma cells. a Semi-quantitative RT-

PCR of GDNF and its receptors GFRa-1, GFRa-2 and RET from

mouse and human cells: astrocytes, NPCs, microglia,fibroblast

(SCRC-1008), GL261 (mouse GL261 glioma cell line), human

glioma cells from two individual patients (Tumor 1 and 2). b-Actin

was used as an internal control for mRNA level. b Secreted GDNF

protein level was measured by ELISA. Bars represent the

mean ± SEM from three independent experiments (*p \ 0.05 com-

pared with fibroblast). c Tissue sections from a mouse inoculated with

mCherry GL261 glioma cells (red) and stained with GDNF (green).

Tissue sections from human cortex (from an epilepsy surgery) as a

control (left panel) and human glioma tissue from HG-glioma patient

(right panel) were stained with GDNF (red) and vimentin (green)

antibody and DAPI (blue)

Acta Neuropathol

123

Author's personal copy

GL261 glioma cells as well as SCRC-1008 fibroblasts

(?vector) versus GDNF overexpressing fibroblasts

(?GDNF) HF were implanted into mouse brains. There

was no significant difference in GFAP intensity and cell

morphology in cells surrounding both siNT and siGDNF

HF (Fig. 5a, b, upper panel) and in cells surrounding

?vector and ?GDNF HF (Fig. 5a, b, lower panel). This

result indicated that neither down-regulation of GDNF in

glioma cells nor over expression of GDNF in non-tumor-

igenic fibroblast cells influenced the glioma-induced

astrogliosis.

GDNF is a potent chemoattractant for microglia

Since we showed that GDNF only affected microgliosis but

not astrogliosis, we aimed to determine the effect of GDNF

on microglial chemotactic behavior. First we tested two

different concentrations (0, 500, and 1,000 ng/ml) of

GDNF on microglial chemotaxis by the transwell assay.

After 24 h incubation, mouse primary microglia migration

into the lower chamber triggered by GDNF increased dose

dependently (0 ng/ml: 35 ± 5 cells/counting field; 500 ng/

ml: 80 ± 5 cells/counting field, and 1,000 ng/ml:

142 ± 12 cells/counting field) (Fig. 6a).

As a second approach, we determined microglia che-

motaxis by analyzing migration into an agarose spot

containing either PBS (as control) or GDNF. The spots

were placed on glass-bottomed Petri dishes and microglial

cells in suspension were subsequently added (Fig. 6b).

Microglial invasion into the GDNF-containing agarose

spots was significantly higher as compared to the control

spots (at 3 h GDNF spot: 422 ± 17 cells/spot; PBS spot:

8 ± 1 cells/spot, p \ 0.001) (Fig. 6b, right panel). To

study the dynamics of microglial invasion, we analyzed

microglial migration by time-lapse recording over 2 h.

Microglia migrated faster and over longer distances into

the GDNF spots as compared to control (p \ 0.001,

Fig. 6c, d). Both velocity and migrated distances increased

in a dose-dependent fashion (p \ 0.001, Fig. 6d).

To verify whether GDNF secreted from GL261 cells

influences microglia migration, the transwell assay was

applied. Conditioned-medium was collected from either

Fig. 4 GDNF-mediated microglia attraction. a GDNF secretion from

non-transfected (GL261), non-targeted siRNA treated GL261 cells

(siNT) and GDNF siRNA treated cells (siGDNF) after 6 days of

transfection was analyzed by ELISA. b After transfection with either

siNT (left panel) or siGDNF (right panel), GL261 cells enclosed in

HF were implanted into the mouse brains (left and right hemisphere,

respectively). After 6 days of implantation, mice were killed and

brain sections were stained with Iba-1 (red) antibody. Cell nuclei

were counterstained with DAPI (blue). Yellow dashed line indicates

the border between HF and brain tissue. c Density of microglia

surrounding HF with siGDNF transfected GL261 cells is shown

(mean ± SEM, n = 11 mice; ***p \ 0.001) compared with control

siNT GL261 cells. d SCRC-1008 fibroblast cells were transfected

with control vector or GDNF. GDNF secretion was detected by

ELISA and shows increased secretion. e After transfection with either

control (left panel) or GDNF vector (right panel), enclosed fibroblast

cells in HF were implanted into mouse brains (left and right

hemisphere, respectively). After 6 days of transplantation, mice were

killed and brain sections were stained with Iba-1 antibody (red). Cell

nuclei were counterstained with DAPI (blue). Yellow dashed lineindicates the border between HF and brain tissue. f Density of

microglia surrounding HF with control (vector) and GDNF over-

expressing fibroblasts is shown (mean ± SEM, n = 5 mice;

***p \ 0.001)

Acta Neuropathol

123

Author's personal copy

siNT or siGDNF GL261 cells and applied to the lower

transwell chamber. During a 24-h transwell migration assay

in control DMEM medium, microglia show basal migration,

while conditioned-medium from siNT-GL261 cells induced

microglia migration. Conditioned medium from siGDNF

GL261 significantly reduced the microglia chemotaxis as

compared to medium from siNT GL261 (Fig. 6e). Based on

these data, we suggest that GDNF released from glioma cells

induces microglia chemoattraction.

Silencing of GDNF by shRNA in GL261 reduced

the tumor size of mice bearing glioma

To determine whether a blockade of GDNF production from

glioma cells suppresses glioma growth in an experimental

glioma model, we stably silenced GDNF in GL261 by shRNA

and studied the influence of GDNF knockdown on tumor

growth. shGDNF GL261 decreased GDNF secretion levels

by 65 % after 1 month of clone selection as determined by

ELISA (Fig. 7a). shNT and shGDNF- GL261 were then

injected into mouse brains. After 2 weeks, tumor size was

visualized and measured by MRI (Fig. 7b, left panel).

Knockdown of GDNF in GL261 cells significantly reduced

glioma tumor size (Fig. 7b, right panel, p \ 0.001). Mouse

brain sections were further stained for Iba-1 and GDNF. Iba-1

positive microglia infiltration was lower when GDNF was

decreased (Fig. 7c). Furthermore, we observed an improved

survival in the group of mice that were inoculated with

shGDNF GL261 (Fig. 7d, p = 0.0074). Altogether, these

in vivo data suggest that GDNF is a crucial factor for glioma

progression.

Discussion

Glioma cells interact with intrinsic brain cells, in particular

microglia/brain macrophages, astrocytes and endogenous

neural precursor cells [4]. Microglial cells are attracted to

the glioma and these tumor-associated brain macrophages

are a mixed cell population of intrinsic microglia and

peripheral monocytes [1]. Depletion of microglia both

in vitro [13] and in vivo [26] impaired tumor growth

indicating that microglia interact with glioma cells and

promote tumor growth [25, 26]. In the present study, we

identified GDNF as a factor released from glioma, which

acts as a chemoattractant for microglia and promotes gli-

oma growth. GDNF was originally identified as a factor

released from a glioma cell line and has been identified as

growth factor acting on different types of neurons [22].

GDNF is also a potent survival factor for dopaminergic

neurons of the nigrostriatal pathway and is considered as a

therapeutic target in Parkinsons disease [16, 19]. It is well

established that GDNF expression is highly upregulated in

glioma [35], and as an autocrine factor, it promotes glioma

cell migration [30]. We confirmed this observation by

staining mouse experimental malignant glioma and human

glioma specimens with anti-GDNF antibody. In the present

study, we furthermore provide evidence that GDNF is a

chemoattractant for microglia. We have demonstrated that

(1) recombinant GDNF is a chemoattractant for cultured

microglial cells as established in two different chemotaxis

assays, (2) glioma-conditioned supernatant from glioma

cells with impaired GDNF expression has a lower potency

as chemoattractant for microglia as compared to control

Fig. 5 GDNF does not alter the

astrogliosis. a Enclosed GL261

cells in HF transfected with

either siNT (upper left panel) or

siGDNF (upper right panel)were implanted into mouse

brains (left and right

hemisphere, respectively).

Enclosed fibroblast transfected

with either control (lower leftpanel) or GDNF (lower rightpanel) vector were implanted

into mouse brains (left and right

hemisphere, respectively). After

6 days, mice were killed and

brain sections were stained for

GFAP (red). Cell nuclei were

counterstained with DAPI

(blue). b The GFAP

fluorescence intensity

surrounding HF did not differ

between the experimental

groups (ns: not significant).

Yellow dashed line indicates the

margin of HF membrane

Acta Neuropathol

123

Author's personal copy

conditioned medium, (3) glioma cells with downregulated

GNDF expression encapsulated into HF and implanted into

mouse brains attract less microglia as compared to control

glioma cells, (4) overexpression of GDNF in a fibroblast

cell line attracted more microglia as HF with control

fibroblasts in our in vivo model, (5) orthotopic implantation

of GDNF knockdown glioma cells into the mouse brain

generates smaller tumors as compared to controls. Impor-

tantly, reduced GDNF levels in gliomas improved survival

in our immunecompetent, orthotopic implantation model.

For our survival data, we also compared unilateral injec-

tions of glioma cells either control or silenced for GDNF

expression. In addition, the unilateral injections resulted in

a significant difference.

Our data also indicate that GDNF is a major but not the

sole chemoattractant for microglia/macrophages to HG-

gliomas. Other factors like MCP-1 (CCL-2), CX3CL1 or

SDF-1 (CXCL-12) may also contribute to microglial

attraction to primary brain tumors. However, the roles of

these factors on the recruitment of microglia towards gli-

oma have only been described in vitro [14, 24, 38].

As a tool to distinguish between cell contact and soluble

factor, we employed the HF approach where glioma cells

were encapsulated and inserted into the brain. In this study,

Fig. 6 GDNF is involved in glioma cell incuced microglia chemo-

taxis. a Microglia was seeded in a transwell plate in the presence of 0,

500, or 1,000 ng/ml of recombinant GDNF for 24 h. Migrated cells

were counted after the membranes were stained with H&E. Barsrepresent the mean ± SEM from three experiments, ***p \ 0.001

and **p \ 0.01 when compared with control. b Microglia was plated

on 35-mm cover slips containing both PBS and GDNF spots. 3 h after

plating, microglia migrated toward GDNF spots (dashed line depicts

border of agarose spot) and cells migrated under/into the spots were

counted (right panel). Bars represent the mean ± SEM from 12

experiments, ***p \ 0.001. c Microglia migration direction was

observed by time-lapse imaging and analyzed by ImageJ software.

Each line represents one cell (25 cells for PBS spot and 27 cells for

GDNF spot). d microglia migration velocity and distance was

calculated. Bars represent the mean ± SEM from 48 experiments,

***p \ 0.001. e Microglia was seeded in a transwell plate in the

presence of control medium (DMEM with 10 % FBS) or glioma-

conditioned medium from non-targeted siRNA transfected GL261

cells (siNT), or glioma-conditioned medium from GDNF-targeted

siRNA transfected GL261 cells (siGDNF). The transwell plate was

then incubated for 24 h. The migrating microglia in the opposite side

of the filter were fixed and stained. Migrated microglia was counted

from at least five fields. Bars represent the mean ± SEM from three

experiments (*p \ 0.05 and ***p \ 0.001)

Acta Neuropathol

123

Author's personal copy

we demonstrated that glioma cells survived and even prolif-

erated in encapsulated HF and that GDNF downregulation

does not affect cell survival or proliferation. Originally, the

strategy of cell encapsulation in hollow fibers was used as a

tool for gene or cell therapy [12, 31]. We adapted and mod-

ified this method to study soluble factors for the interaction of

glia and glioma. When compared to the conventional non-

encapsulated cell implantation methods [15], the HF

approach allowed us to distinguish between the influence of

secreted factors and contact-mediated interaction and

allowed us to test the impact of human glioma cells on mi-

croglial attraction and astrocyte activation in immune

competent animals. We compared the impact of HF with

glioma cells versus HF containing fibroblasts. We found that

the glioma cells resulted in a significantly higher accumula-

tion of microglial cells. Also for the glioma cells injected

directly into brain, we found a lower density of microglia in

the GDNF-silenced glioma cells. We assume that the reduced

density of microglia results in a reduced tumor growth as we

had shown in our previous publications [25–27] Glioma cells

also induce reactive astrogliosis [20] accompanied by an

upregulation of GFAP and a morphological transformation.

We observed that glioma implanted fibers induce astrogliosis

as well. This effect is specific for glioma cells since implanted

fibroblast did not trigger such a response. We found, however,

that GDNF depletion or overexpression does not affect as-

trogliosis indicating that factors other than GDNF-mediated

glioma–astrocyte interaction.

In summary, we identified GDNF as a HG-glioma

released factor that specifically leads to microglial attrac-

tion. When we reduce GDNF in glioma cells, the microglia

attraction is affected and tumor growth is diminished.

Thus, our study supports the idea that microglia play an

important role in tumor growth, invasion and progression

and thus can become a novel target for therapeutic

strategies.

Fig. 7 GDNF knockdown decrease glioma tumor size and prolong

survival. a GDNF knockdown in GL261 cells was accomplished by

transfecting GDNF shRNA construct and measuring GDNF secretion

after 1 month of selection (*p \ 0.05). b Coronal T2-weighted in vivo

MR image of the mouse brain depicting intracranial HG-gliomas

(borders highlighted by dashed lines). Note the difference in tumor

size between the control GL261 (shNT, left hemisphere) and the

GDNF knockdown GL261 (shGDNF, right hemisphere). Tumor size

was calculated by manual segmentation of the tumor borders on the

MR images (n = 10 mice, ***p \ 0.001). c After MRI, mice were

sacrificed and brain was cryosectioned and stained with GDNF

(magenta) and Iba-1 (green) antibody. d Kaplan–Meier survival

curves of glioma cell implanted mice. Survival of mice was tracked

following inoculation of 2 9 104 shNT or shGDNF GL261 cells

(n = 6 mice each group, p = 0.0074)

Acta Neuropathol

123

Author's personal copy

Acknowledgments This work was supported by the graduate school

of NeuroCure at the Charite, Berlin, (stipend to M. C. Ku) and by

Deutsche Forschungsgemeinschaft (TR 43). We are thankful to Prof.

Carlos Ibanez for providing GDNF cDNA construct for transfection

and Prof. Jochen Meier for providing human brain tissue. We thank

Dr. Zoltan Cseresnyes and Dr. Anje Sporbert for technical assistance

with confocal microscopy. We appreciate the support of Babette

Dieringer for MR imaging and Maria Pannell for manuscript proof

reading.

Conflict of interest The authors declare that they have no conflict

of interest.

References

1. Badie B, Schartner J (2001) Role of microglia in glioma biology.

Microsc Res Tech 54(2):106–113. doi:10.1002/jemt.112510.

1002/jemt.1125

2. Badie B, Schartner J, Klaver J, Vorpahl J (1999) In vitro mod-

ulation of microglia motility by glioma cells is mediated by

hepatocyte growth factor/scatter factor. Neurosurgery 44(5):

1077–1082 (discussion 1082–1073)

3. Broadhead KW, Biran R, Tresco PA (2002) Hollow fiber mem-

brane diffusive permeability regulates encapsulated cell line

biomass, proliferation, and small molecule release. Biomaterials

23(24):4689–4699. doi:10.1016/S0142-9612(02)00212-0

4. Charles NA, Holland EC, Gilbertson R, Glass R, Kettenmann H

(2011) The brain tumor microenvironment. Glia 59(8):1169–

1180. doi:10.1002/glia.21136

5. Cornejo M, Nambi D, Walheim C, Somerville M, Walker J, Kim

L, Ollison L, Diamante G, Vyawahare S, de Bellard ME (2010)

Effect of NRG1, GDNF, EGF and NGF in the migration of a

Schwann cell precursor line. Neurochem Res 35(10):1643–1651.

doi:10.1007/s11064-010-0225-0

6. Dhandapani KM, Khan MM, Wade FM, Wakade C, Mahesh VB,

Brann DW (2007) Induction of transforming growth factor-beta1

by basic fibroblast growth factor in rat C6 glioma cells and

astrocytes is mediated by MEK/ERK signaling and AP-1 acti-

vation. J Neurosci Res 85(5):1033–1045. doi:10.1002/jnr.21182

7. Dudanova I, Gatto G, Klein R (2010) GDNF acts as a chemo-

attractant to support ephrinA-induced repulsion of limb motor

axons. Curr Biol 20(23):2150–2156. doi:10.1016/j.cub.2010.11.

021

8. Edwards LA, Woolard K, Son MJ, Li A, Lee J, Ene C, Mantey

SA, Maric D, Song H, Belova G, Jensen RT, Zhang W, Fine HA

(2011) Effect of brain- and tumor-derived connective tissue

growth factor on glioma invasion. J Natl Cancer Inst

103(15):1162–1178. doi:djr22410.1093/jnci/djr224

9. Franklin K, Paxinos G (2007) The mouse brain in stereotaxic

coordinates, 3rd edn. Academic Press, San Diego

10. Geranmayeh F, Scheithauer BW, Spitzer C, Meyer FB, Svensson-

Engwall AC, Graeber MB (2007) Microglia in gemistocytic

astrocytomas. Neurosurgery 60(1):159–166 (discussion 166). doi:

10.1227/01.NEU.0000249192.30786.67

11. Glass R, Synowitz M, Kronenberg G, Walzlein JH, Markovic DS,

Wang LP, Gast D, Kiwit J, Kempermann G, Kettenmann H (2005)

Glioblastoma-induced attraction of endogenous neural precursor

cells is associated with improved survival. J Neurosci 25(10):2637–

2646. doi:25/10/263710.1523/JNEUROSCI.5118-04.2005

12. Han Q, Sun W, Lin H, Zhao W, Gao Y, Zhao Y, Chen B, Xiao Z,

Hu W, Li Y, Yang B, Dai J (2009) Linear ordered collagen

scaffolds loaded with collagen-binding brain-derived neurotro-

phic factor improve the recovery of spinal cord injury in rats.

Tissue Eng Part A 15(10):2927–2935. doi:10.1089/ten.TEA.2008.

0506

13. Hussain SF, Yang D, Suki D, Aldape K, Grimm E, Heimberger

AB (2006) The role of human glioma-infiltrating microglia/

macrophages in mediating antitumor immune responses. Neuro

Oncol 8(3):261–279. doi:10.1215/15228517-2006-008

14. Kenig S, Alonso MB, Mueller MM, Lah TT (2009) Glioblastoma

and endothelial cells cross-talk, mediated by SDF-1, enhances

tumour invasion and endothelial proliferation by increasing

expression of cathepsins B, S, and MMP-9. Cancer Lett 289(1):53–

61. doi:10.1016/j.canlet.2009.07.014

15. Kim YT, Hitchcock R, Broadhead KW, Messina DJ, Tresco PA

(2005) A cell encapsulation device for studying soluble factor

release from cells transplanted in the rat brain. J Control Release

102(1):101–111. doi:10.1016/j.jconrel.2004.10.003

16. Kirik D, Georgievska B, Bjorklund A (2004) Localized striatal

delivery of GDNF as a treatment for Parkinson disease. Nat

Neurosci 7(2):105–110. doi:10.1038/nn1175nn1175

17. Koelsch A, Feng Y, Fink DJ, Mata M (2010) Transgene-mediated

GDNF expression enhances synaptic connectivity and GABA

transmission to improve functional outcome after spinal cord

contusion. J Neurochem 113(1):143–152. doi:10.1111/j.1471-

4159.2010.06593.x

18. Komohara Y, Ohnishi K, Kuratsu J, Takeya M (2008) Possible

involvement of the M2 anti-inflammatory macrophage phenotype

in growth of human gliomas. J Pathol 216(1):15–24. doi:10.1002/

path.2370

19. Lang AE, Gill S, Patel NK, Lozano A, Nutt JG, Penn R, Brooks

DJ, Hotton G, Moro E, Heywood P, Brodsky MA, Burchiel K,

Kelly P, Dalvi A, Scott B, Stacy M, Turner D, Wooten VG, Elias

WJ, Laws ER, Dhawan V, Stoessl AJ, Matcham J, Coffey RJ,

Traub M (2006) Randomized controlled trial of intraputamenal

glial cell line-derived neurotrophic factor infusion in Parkinson

disease. Ann Neurol 59(3):459–466. doi:10.1002/ana.20737

20. Le DM, Besson A, Fogg DK, Choi KS, Waisman DM, Goodyer

CG, Rewcastle B, Yong VW (2003) Exploitation of astrocytes by

glioma cells to facilitate invasiveness: a mechanism involving

matrix metalloproteinase-2 and the urokinase-type plasminogen

activator-plasmin cascade. J Neurosci 23(10):4034–4043

21. Leung SY, Wong MP, Chung LP, Chan AS, Yuen ST (1997)

Monocyte chemoattractant protein-1 expression and macrophage

infiltration in gliomas. Acta Neuropathol 93(5):518–527

22. Lin LF, Doherty DH, Lile JD, Bektesh S, Collins F (1993)

GDNF: a glial cell line-derived neurotrophic factor for midbrain

dopaminergic neurons. Science 260(5111):1130–1132

23. Lu DY, Leung YM, Cheung CW, Chen YR, Wong KL (2010) Glial

cell line-derived neurotrophic factor induces cell migration and

matrix metalloproteinase-13 expression in glioma cells. Biochem

Pharmacol 80(8):1201–1209. doi:10.1016/j.bcp.2010.06.046

24. Magge SN, Malik SZ, Royo NC, Chen HI, Yu L, Snyder EY,

O’Rourke DM, Watson DJ (2009) Role of monocyte chemoattrac-

tant protein-1 (MCP-1/CCL2) in migration of neural progenitor

cells toward glial tumors. J Neurosci Res 87(7):1547–1555. doi:

10.1002/jnr.21983

25. Markovic DS, Glass R, Synowitz M, Rooijen N, Kettenmann H

(2005) Microglia stimulate the invasiveness of glioma cells by

increasing the activity of metalloprotease-2. J Neuropathol Exp

Neurol 64(9):754–762. doi:00005072-200509000-00002

26. Markovic DS, Vinnakota K, Chirasani S, Synowitz M, Raguet H,

Stock K, Sliwa M, Lehmann S, Kalin R, van Rooijen N, HolmbeckK, Heppner FL, Kiwit J, Matyash V, Lehnardt S, Kaminska B, Glass

R, Kettenmann H (2009) Gliomas induce and exploit microglial

MT1-MMP expression for tumor expansion. Proc Natl Acad Sci

USA 106(30):12530–12535. doi:10.1073/pnas.0804273106

27. Markovic DS, Vinnakota K, van Rooijen N, Kiwit J, Synowitz M,

Glass R, Kettenmann H (2011) Minocycline reduces glioma

Acta Neuropathol

123

Author's personal copy

expansion and invasion by attenuating microglial MT1-MMP

expression. Brain Behav Immun 25(4):624–628. doi:10.1016/j.bbi.

2011.01.015

28. Okada M, Saio M, Kito Y, Ohe N, Yano H, Yoshimura S, Iwama

T, Takami T (2009) Tumor-associated macrophage/microglia

infiltration in human gliomas is correlated with MCP-3, but not

MCP-1. Int J Oncol 34(6):1621–1627

29. Platten M, Kretz A, Naumann U, Aulwurm S, Egashira K, Is-

enmann S, Weller M (2003) Monocyte chemoattractant protein-1

increases microglial infiltration and aggressiveness of gliomas.

Ann Neurol 54(3):388–392. doi:10.1002/ana.10679

30. Song H, Moon A (2006) Glial cell-derived neurotrophic factor

(GDNF) promotes low-grade Hs683 glioma cell migration

through JNK, ERK-1/2 and p38 MAPK signaling pathways.

Neurosci Res 56(1):29–38. doi:10.1016/j.neures.2006.04.019

31. Visted T, Bjerkvig R, Enger PO (2001) Cell encapsulation

technology as a therapeutic strategy for CNS malignancies.

Neuro Oncol 3(3):201–210

32. Voisin P, Bouchaud V, Merle M, Diolez P, Duffy L, Flint K,

Franconi JM, Bouzier-Sore AK (2010) Microglia in close vicinity of

glioma cells: correlation between phenotype and metabolic altera-

tions. Front Neuroenergetics 2:131. doi:10.3389/fnene.2010.00131

33. Wan G, Too HP (2010) A specific isoform of glial cell line-

derived neurotrophic factor family receptor alpha 1 regulates

RhoA expression and glioma cell migration. J Neurochem

115(3):759–770. doi:10.1111/j.1471-4159.2010.06975.x

34. Watters JJ, Schartner JM, Badie B (2005) Microglia function in

brain tumors. J Neurosci Res 81(3):447–455. doi:10.1002/jnr.20485

35. Wiesenhofer B, Stockhammer G, Kostron H, Maier H, Hinterh-

uber H, Humpel C (2000) Glial cell line-derived neurotrophic

factor (GDNF) and its receptor (GFR-alpha 1) are strongly

expressed in human gliomas. Acta Neuropathol 99(2):131–137

36. Wiggins H, Rappoport J (2010) An agarose spot assay for chemotactic

invasion. Biotechniques 48(2):121–124. doi:10.2144/000113353

37. Zhang GJ, Chen TB, Hargreaves R, Sur C, Williams DL Jr (2008)

Bioluminescence imaging of hollow fibers in living animals: its

application in monitoring molecular pathways. Nat Protoc

3(5):891–899. doi:10.1038/nprot.2008.52

38. Zhao D, Najbauer J, Garcia E, Metz MZ, Gutova M, Glackin CA,

Kim SU, Aboody KS (2008) Neural stem cell tropism to glioma:

critical role of tumor hypoxia. Mol Cancer Res 6(12):1819–1829.

doi:10.1158/1541-7786.MCR-08-0146

Acta Neuropathol

123

Author's personal copy