University of ZurichZurich Open Repository and Archive

Winterthurerstr. 190

CH-8057 Zurich

http://www.zora.uzh.ch

Year: 2008

Expression and putative functions of GDF-15, a member of theTGF-β superfamily, in human glioma and glioblastoma cell lines

Strelau, J; Schmeer, C; Peterziel, H; Sackmann, T; Herold-Mende, C; Steinert, H;Weller, M; Unsicker, K

Strelau, J; Schmeer, C; Peterziel, H; Sackmann, T; Herold-Mende, C; Steinert, H; Weller, M; Unsicker, K (2008).Expression and putative functions of GDF-15, a member of the TGF-β superfamily, in human glioma andglioblastoma cell lines. Cancer Letters, 270(1):30-39.Postprint available at:http://www.zora.uzh.ch

Posted at the Zurich Open Repository and Archive, University of Zurich.http://www.zora.uzh.ch

Originally published at:Cancer Letters 2008, 270(1):30-39.

Strelau, J; Schmeer, C; Peterziel, H; Sackmann, T; Herold-Mende, C; Steinert, H; Weller, M; Unsicker, K (2008).Expression and putative functions of GDF-15, a member of the TGF-β superfamily, in human glioma andglioblastoma cell lines. Cancer Letters, 270(1):30-39.Postprint available at:http://www.zora.uzh.ch

Posted at the Zurich Open Repository and Archive, University of Zurich.http://www.zora.uzh.ch

Originally published at:Cancer Letters 2008, 270(1):30-39.

Expression and putative functions of GDF-15, a member of theTGF-β superfamily, in human glioma and glioblastoma cell lines

Abstract

Recent studies have demonstrated growth-inhibiting effects of growth differentiation factor-15(GDF-15) on different cancer cell lines invitro and on tumor growth in vivo. Here, we present dataconcerning expression of GDF-15 in glioblastoma. We found low levels of GDF-15 transcripts inprimary glioblastoma. Thus, GDF-15 expression might be exploited as a useful indicator fordistinguishing primary from other glial derived tumors. In contrast to the documented proapoptotic andanti-tumorigenic activities of GDF-15 in several cancer cell lines, our data suggest that GDF-15 does notdecrease proliferation of glioblastoma cell lines, while its effects on invasiveness are not consistent.

Elsevier Editorial System(tm) for Cancer Letters

Manuscript Draft

Manuscript Number: CAN-D-07-01136R1

Title: EXPRESSION AND PUTATIVE FUNCTIONS OF GDF-15, A MEMBER OF THE TGF-ß

SUPERFAMILY, IN HUMAN GLIOMA AND GLIOBLASTOMA CELL LINES

Article Type: Original Research Paper

Section/Category:

Keywords: GDF-15; TGF-ß superfamily; glioblastomas; tumor

Corresponding Author: Dr. Jens Strelau, Ph.D.

Corresponding Author's Institution:

First Author: Jens Strelau, Dr.

Order of Authors: Jens Strelau, Dr.; Corina Schmeer, Dr.; Heike Peterziel, Dr.; Tina Paech; Christel Herold-

Mende; Hans H Steiner, Prof. Dr. ; Michael Weller, Prof. Dr.; Klaus Unsicker, Prof. Dr.

Manuscript Region of Origin:

Response to Reviewers: Response to Reviewers:

Reviewer 1

1. First this subtitle "GDF-15 is differentially expressed in cultures of primary and secondary glioblastoma" is

misleading.

We changed the subtitle into “Low expression of GDF-15 in cultures of primary glioblastomas” and present

in the subsequent chapter data on diffuse astrocytomas grade II, anaplastic astrocytomas grade III,

secondary glioblastomas grade IV, primary glioblastomas grade IV, oligodendroglioblastomas grade II and

III, and oligoastrocytomas grade II and III.

We also changed the first subtitle of the Results part into “GDF-15 immunoreactivity in de novo glioblastoma,

high grade astrocytoma and secondary glioblastoma” and subdivided figure 1 into A and B for the sake of

clarity of the immunohistochemical results.

2. No details are given about how these cultures were derived, maintained or what passage numbers were

used in the present studies.

Details are now provided in the Materials and Methods section on how the cultures were derived.

Furthermore, data on GDF-15 expression in oligodendrogliomas grade II and III and oligoastrocytomas

grade II and III were added to figure 2 and Results. We also incorporated a table comprising tumor type,

passage numbers, age and sex of each specimen (Tab.1).

4. Did the authors conduct similar studies on the primary cell lines used in Figure 2?

Figure 2 shows real-time PCR results with cDNA of 59 different primary cell cultures with low passage

numbers provided by the Molecular Biology Laboratories, Department of Neurosurgery, University of

Heidelberg. Since cell cultures with similarly low passage numbers were not available we did not perform

invasion-, proliferation-, and TUNEL-assays on primary cell lines.

5. How do the authors explain the differences in responsiveness of the different cell lines?

Our studies show that the GDF-15 antiserum decreases proliferation in 8 of 9 cell cultures. Exogenous GDF-

15 increased cell division in 4 cell lines, while the remaining 5 cell lines did not respond. Several possibilities

may be conceived to explain this result. First, cell lines unresponsive to exogenous GDF-15, but responsive

to the anti-GDF-15 might already be maximally stimulated by the endogenous GDF-15. Moreover, non-

responsive cell lines might require, in addition to GDF-15, a synergistically acting factor, which was not

available at the required amounts. Alternatively, GDF-15 receptors in non-responsive cell lines might already

be maximally saturated with endogenous GDF-15. A similar explanation might account for the lack of

response to exogenous GDF-15 in the invasion experiments. It should be noted, however, that both U87MG

and, to a lesser extent, LN-428 and LN-229, showed a tendency to respond to GDF-15.

It has been shown that, dependent on tumor type and origin (for review see Eling at al., 2006; Bauskin et al.,

2006), GDF-15 can exhibit either tumorigenic or anti-tumorigenic activity. The cell lines investigated in our

studies are derived from tumors of different stages and origin, and it is therefore conceivable that GDF-15

may regulate in either a positive or negative manner, depending on molecular and cellular contexts.

Antagonistic effects have also been shown for TGF � and its effects on different glioma cell lines (see also

Jennings et al., 1991).

Although previous studies suggested that GDF-15 can regulate invasiveness in cancer cell lines, results of

these former studies were based on a single cell line, without comparing the results with other cell lines

derived from the same cancer types. It has been shown, for example, that GDF-15 significantly increases

invasiveness of SNU-216, a human gastric cancer cell line (Lee et al., 2003). Evidently, this does not prove

that all available gastric cancer cell lines would respond in the same way. Our study compares GDF-15

dependent invasion in 8 different cancer cell lines, which may differ in a variety of subtle respects.

6. Lastly, no effect on apoptosis was noted. How do the authors explain this result, which is different from

previously published studies?

Most in vitro studies suggest GDF-15- induces apoptosis in cancer cell lines. However, data from the

literature provide high expression of GDF-15 in many tumors, a finding that is hard to be reconciled with the

proposed tumor suppressor activity of GDF-15. Previous studies from our laboratory clearly demonstrate

that GDF-15 can protect cerebellar granule cells from apoptotic cell death in vitro (Subramaniam et al.,

2003), i.e. GDF-15 may function as an anti-apoptotic factor. To our knowledge, none of the previously

published studies has shown pro-apoptotic effects of GDF-15 in glioma cultures.

7. The conclusions of this study are overstated.

Conclusions have been changed

Reviewer 2

1. How are primary and secondary GBM differentiated (clinically or molecular).

Primary and secondary GBM were differentiated as described earlier [22] In brief, tumors were

histopathologically classified according to the classification of the WHO. Tumors were characterized as

secondary GBM in patients with previous identified low grade (WHO II and WHO III) glioblastomas in

accordance with Kleihues.

2. Why was no fresh material used for qPCR?

We have now included fresh material provided by the Molecular Biology Laboratories, Department of

Neurosurgery, University of Heidelberg. We got 9 frozen tissue samples of primary GBM and used the

samples for qPCR and immunohistochemistry. The results of the qPCR with fresh primary GBM confirmed

data obtained by IHC (low GDF-15 expression in 8 out of 9 tumors, (see also the figure on the last page of

our revision notes). Three of the primary cultures used for qPCR correspond to fresh tissue samples (Tab.1).

Since one secondary GBM, two anaplastic astrocytoma and one oligodendroblastoma were provided as

frozen sections on coverslips we could not perform qPCR and used the material directly for

immunohistochemistry. However, primary cells derived from these tumors revealed high amounts of GDF-15

mRNA (Tab.1), although we were not able to show any ir in the oligodendroglioma grade III.

3. What is a low-passage culture and how was the immunohisto / qPCR correlation?

We added a list with tumor number, passage number, sex and age of the total material used in this study

(Tab. 1).

4. Table 2 should be presented earlier in the results section.

The correlation of p53 and GDF-15 is now mentioned in the Results section.

5. Discussion should be shortened.

We now shortened the discussion.

Literature:

Asne R. Bauskin et al., Role of Macrophage Inhibitory Cytokine-1 in Tumorigenesis and Diagnosis of Cancer.

Cancer Res. 2006; 66: (10). May 15, 4983-4986

Thomas E. Eling et al., NSAID Activated Gene (NAG-1), a Modulator of Tumorigenesis. J Biochem. and Mol.

Biol. 2006; 39: (6). Nov. 649-655

Mark T. Jennings et al., TGF-ß1 and TGF-ß2 are potential growth regulators for low-grade and malignant

gliomas in vitro: Evidence in support of an autocrine hypothesis. Int. J. Cancer 1991; 49, 129-139)

1

Cancer Letters

EXPRESSION AND PUTATIVE FUNCTIONS OF GDF-15, A MEMBER

OF THE TGF-ß SUPERFAMILY, IN HUMAN GLIOMA AND

GLIOBLASTOMA CELL LINES

Jens Strelau 1, Corina Schmeer 1, Heike Peterziel 1, Tina Paech 1, Christel Herold-

Mende 2, Hans Steiner 3, Michael Weller 4 and Klaus Unsicker 1

1Department of Neuroanatomy and Interdisciplinary Center for Neurosciences (IZN)

University of Heidelberg, Germany2 Molecular Biology Laboratories, Department of Neurosurgery, University of Heidelberg,

Germany 3 Department of Neurosurgery, Nürnberg Süd, Nürnberg, Germany4 Department of Neurology, University Hospital Zurich, Switzerland.

Address for correspondence:

Dr. Jens Strelau

Neuroanatomy & Interdisciplinary Center for Neurosciences (IZN),

University of Heidelberg,

INF 307

D-69120 Heidelberg, Germany

Fax: 49-6221-54 56 04

e-mail: jens.strelau@urz.uni-heidelberg.de

Manuscript

2

Abstract

Recent studies have demonstrated growth-inhibiting effects of Growth differentiation factor-

15 (GDF-15) on different cancer cell lines in vitro and on tumor growth in vivo. Here, we

present data concerning expression of GDF-15 in glioblastoma. We found low levels of GDF-

15 transcripts in primary glioblastoma. Thus, GDF-15 expression might be exploited as a

useful indicator for distinguishing primary from other glial derived tumors. In contrast to the

documented proapoptotic and anti-tumorigenic activities of GDF-15 in several cancer cell

lines, our data suggest that GDF-15 does not decrease proliferation of glioblastoma cell lines,

while its effects on invasiveness are not consistent.

Key words: GDF-15, TGF- superfamily, glioblastomas, tumor

3

Introduction

Growth differentiation factor-15 (GDF-15), also known as macrophage inhibitory cytokine-1

[1], prostate-derived factor [2], placental bone morphogenetic protein [3], and nonsteroidal

anti-inflammatory drug-activated protein-1 [4], is a divergent member of the TGF-ß

superfamily and widely expressed in epithelial tissues and in brain [5, 6, 7].

Most recently, several reports have provided evidence for functional links between GDF-15

and factors involved in tumorigenesis. The tumor suppressor gene product p53, for instance,

strongly induces GDF-15, indicating for the first time a functional connection between p53

and the TGF-ß superfamily [8-12]. Nonsteroidal anti-inflammatory drugs (NSAIDs) that

inhibit tumor development can also increase GDF-15 expression [13]. Other studies have

documented increased levels of GDF-15 in serum and in tumors of patients with metastatic

cancer [14-18]. Perhaps most excitingly, GDF-15 has been reported to exert proapoptotic and

anti-tumorigenic functions on human colorectal, prostate, breast, and mammary cancer cells

in vitro and on colon and glioblastoma tumors in vivo [4, 8, 19-21]. Together, these findings

strongly suggest a putative role of GDF-15 in tumor development and/or progression and

underscore the importance of further exploration of an involvement of GDF-15 in

tumorigenesis.

We report here the expression of GDF-15 in human glioblastoma tissues, glioblastoma

primary cultures, and glioblastoma cell lines. Our analyses reveal that GDF-15 may be

implicated in the regulation of proliferation and invasiveness of human glioblastoma cells.

4

Materials and Methods

Samples and Cell culture

Tissue specimens were obtained intraoperatively from central parts of the tumors. Primary

and secondary GBM were histopathologically classified according to the classification of the

WHO. Tumors were characterized as secondary GBM in patients with previous identified low

grade (WHO II and WHO III) glioblastomas. Primary cultures of glioma tissues (Tab. 1) were

established as described earlier [22]. The human glioma cell lines A172, D247MG, LN-18,

LN-229, LN-308, LN-319, and LN-428 (see Tab. 2) were maintained in DMEM with 4.5 g

glucose/L supplemented with 10% fetal bovine serum (FBS), penicillin and streptomycin.

Cultures were maintained in a humidified 5% CO2 atmosphere at 37oC.

GDF-15 immunohistochemistry

Samples of primary and secondary glioblastomas were obtained from surgical cases in the

Department of Neurosurgery, University of Heidelberg, Germany. Snap frozen samples were

embedded in Tissue-Tek (Sakura, Ramsey, MI, USA) and directly cut into 16µm slices. For

immunohistochemistry, sections were fixed in 4% PFA for 1 hour. Non-specific peroxidase

activity was blocked by incubating sections in 1% hydrogen peroxide in PBS for 10 min at

room temperature (RT). Non-specific antibody binding was blocked using blocking buffer

(5% goat serum and 0.1% Triton X-100 in PBS) for 1 hour at RT. After washing, coverslips

were incubated with rabbit anti-mouse GDF-15 polyclonal antibodies (affinity-purified and

diluted 1:200) or with PBS overnight at 4oC. Slides were subsequently washed 3 times for 5

min in PBS and incubated 30 min at RT with biotinylated goat anti-rabbit IgG (Vectastain

Elite ABC Kit, Vector Laboratories, Burlingame, CA, USA). All subsequent steps were

performed according to the manufacturer’s instructions. Slides were counterstained with

hematoxylin, coverslipped, and photographed using a Zeiss Axioplan 2 equipped with

5

AxioVision 3.1 software. Immunofluorescence was performed as described [23]. Primary

antibodies used were affinity purified polyclonal anti-GDF-15, monoclonal anti-CD68 (BD

Biosciences, San Jose, CA, USA), and monoclonal anti-GFAP (Sigma, Chicago, IL, USA).

Secondary antibodies were anti-mouse IgG Cy2 and anti-rabbit IgG Cy3 (Jackson

ImmunoResearch, West Baltimore Pike West Grove, PA, USA)

Gel electrophoresis and immunoblot analyses

Protein extracts were prepared by homogenizing the cells in electrophoresis sample buffer,

and the protein content was determined using densitometry [24]. 25µg protein extract per lane

were loaded on SDS-polyacrylamide gels and transferred to nitrocellulose membranes

(Hybond ECL, Amersham Pharmacia, Göttingen, Germany) by electroblotting. The

membranes were incubated with purified polyclonal rabbit anti-rat GDF-15/MIC-1 antibody

for 16 h at 4 °C. Bound antibody was detected with a peroxidase-conjugated secondary

antibody and the ECL western blotting substrate system (Amersham Pharmacia) according to

the manufacturer´s manual. Samples of purified recombinant GDF-15 were visualized with

Coomassie blue and quantified by densitometry comparison [24] with defined concentrations

of protein standards.

ELISA

For the determination of secreted GDF-15 from the glioma cell lines, 25.000 cells/well were

seeded in 48 well plates. The cell culture supernatants were collected after 24 hours and the

amount of GDF-15 determined by sandwich ELISA using a monoclonal anti-human GDF-15

capture antibody (R&D Systems, MAB957) and a biotinylated goat anti-human GDF-15

detection antibody (R&D Systems, BAF940) according to the manufacturer’s protocol.

Recombinant human GDF-15 (R&D Systems) diluted in medium (15 pg/ml to 250 pg/ml)

6

was used as a standard for the determination of the total amount of GDF-15 secreted by the

cells. The number of cells per well after 14 hours was determined by direct counting. The

experiment was performed three times with every cell line tested in triplicate per experiment.

Total RNA extraction and cDNA synthesis

Samples of tumor tissue (100 – 200 mg each) and cultured glioma cells were lysed in

peqGOLD TriFast reagent (Peqlab, Erlangen, Germany), and phenol-chloroform extraction of

total RNA was performed. Prior to reverse transcription RNA samples were treated with RQ1

DNase (Promega). cDNA synthesis was then performed using M-MLV Reverse Transcriptase

(Promega). Additionally, total RNA was extracted from primary (low-passage) cultures

established from glioma tissue of different malignancy grades.

Real-time quantitative RT-PCR analysis

Relative quantitation of human GDF-15 and GAPDH transcripts in cDNA samples was

performed on an ABI PRISM 7000 Sequence Detection System using the corresponding

TaqMan Assays-on-Demand Gene Expression Products (Applied Biosystems) and following

the manufacturer’s protocol. Each sample was analyzed in triplicate. GAPDH expression was

considered as an internal control to which GDF-15 expression was normalized.

Cell proliferation assays

Cells were seeded onto 96-well plates at a density of 104 (for U373MG: 5 x 103) cells/well.

On the next day, cultures were treated with rabbit anti-mouse GDF-15 polyclonal serum (5

L/ml), pre-immune serum (5 l/ml) as a control, or human recombinant GDF-15 (20 ng/ml)

(R&D). 24 hours later, cell proliferation was measured using 5-Bromo-2`-deoxy-uridine

7

labeling and the Detection Kit III (Roche Applied Science), following the manufacturer’s

instructions.

In vitro cell invasion assays

The ability of cells to migrate through Matrigel-coated filters was determined using the 24-

well plate BD BioCoat Matrigel Invasion Chamber (BD Biosciences, Bedford, MA, USA).

Briefly, cells were seeded at a density of 2.5x104 cells in 500 L DMEM onto Matrigel

inserts. Rabbit anti-mouse GDF-15 polyclonal serum (5 l/m), pre-immune serum (5 l/ml, as

a control), or human recombinant GDF-15 (20 ng/ml) were added to the inserts. Wells of the

24-well plates contained DMEM supplemented with 5% FBS as a chemoattractant. After

incubation for 24 h at 37oC in 5% CO2 the cells that had not invaded the filters were removed

with cotton swabs, and cells that had migrated to the lower surface of the filter were fixed in

methanol. Cells were then stained with Toluidine Blue (Merck, Darmstadt, Germany) and

counted in 32 fields (x200) per filter, covering the periphery and the center of the membrane.

All experiments were performed in triplicate.

TUNEL

Cells were washed and fixed with 4% PFA, permeabilized with 0.2% Triton X-100, and

processed for TUNEL (Promega). 5 µg/ml DAPI was added to stain cell nuclei.

Photomicrographs from 4-6 different fields in each coverslip were captured under green

channels and merged using adobe photoshop 5.5. Typically, about 1000 cells were analyzed

for the number of TUNEL-positive glioblastoma cells.

8

Statistical analysis

Data were expressed as means ± SEM. All experiments were performed in triplicates or

duplicates and repeated at least three times. Statistical analyses among groups were performed

using the t test (Origin 6.1). P-values are as follows: * P < 0.05, ** P < 0.01, and *** P <

0.001.

9

Results

GDF-15 immunoreactivity in de novo glioblastoma, high grade astrocytoma and

secondary glioblastoma

We first investigated the possible presence of GDF-15 protein in surgically removed human

primary and secondary glioblastomas using immunohistochemistry. Only one out of the nine

investigated primary glioblastomas grade IV showed moderate GDF-15 immunoreactivity (ir)

(Figure 1A; left panel). GDF-15 staining was localized to tumor cells, but absent from blood

vessels. The right panel shows the control experiment where the GDF-15 antibody was

omitted. In contrast to the largely GDF-15 negative primary glioblastomas we found

prominent staining for GDF-15 ir in 2 anaplastic astrocytomas grade III (AAG III) and one

secondary glioblastoma. The upper left panel of Figure 1B exemplifies GDF-15 ir in one of

the AAG III. Notably, the stroma and perivascular spaces of the AAG III were

immunoreactive for GDF-15. An oligodendroglioma grade III (OG III) was negative for

GDF-15 ir. In samples of tumor tissues, where GDF-15 ir was detectable, there was no

costaining of GDF-15 with glial fibrillary acidic protein (GFAP; Figure 1B; bottom left

panel). Similarly, GDF-15 immunoreactive cells in AAG III (Figure 1B; bottom right panel)

and secondary glioblastomas did not show co-labeling for the microglial marker CD68.

Together, these data suggest the presence of GDF-15 protein in GFAP- and CD68-negative

cells, preferentially in AAG III and secondary glioblastoma.

Lowest expression of GDF-15 in cultures of primary glioblastomas

Diffuse astrocytomas grade II (DAG II), anaplastic astrocytomas grade III (AAG III), and

secondary glioblastomas grade IV (SGG IV) represent distinct stages in the evolution of

secondary glioblastomas, whereas primary glioblastomas grade IV (PGG IV) develop along a

different genetic pathway [25]. Other brain tumours distinguishable from those mentioned

10

above due to their distinct genetical alteration, mainly characterized by a combined loss of

chromosomes 1p and 19q, are oligodendroglioblastomas and oligoastrocytomas grade II and

III (OGG II, OGG III; OAG II, OAG III) [26,27]. Figure 2 shows the results of quantitative

real-time PCR determinations of GDF-15 mRNA expressed in cultures derived from 8 DAG

II, 8 AAG III, 8 SGG IV, 8 PGG IV, 5 OGG II, 5 OGG III, 5 OAG II, and 5 OAG III. High

and largely comparable expression levels were found in DAG II, AAG III, and SGG IV

(Figure 2). In contrast, PGG IV showed significantly lower levels of GDF-15 expression. This

difference was even more pronounced when GDF-15 expression in PGG IV was compared to

the average expression levels of cell cultures derived from oligodendroglioblastomas grade II

(OGG II) and anaplastic oligodendroglioblastomas grade III (OGG III). The most frequent

type of mixed glioblastomas, OAG II, also showed a significantly higher GDF-15 level

compared to PGG IV. Thus, differences in GDF-15 expression might be exploited as a useful

indicator for distinguishing primary from other glial derived tumors.

GDF-15 expression does not correlate with spontaneous proliferation or the p53 status

of glioma cell lines

In an attempt to elucidate putative functions of GDF-15 on glioblastoma growth we

investigated a panel of 9 human glioma cell lines (LN-319, U87MG, A172, D-247MG, LN-

229, LN-308, LN-428, LN-18 and U373MG). We first compared the expression levels of

endogenous GDF-15 mRNA in relation to the spontaneous proliferation rate for each human

cell line. Figure 3A shows the wide range of mRNA expression levels in these cell lines, from

very high rates in A172 and D-247MG cells to up to 15-fold lower levels in LN-18 and

U87MG cells. Similar differences were found in terms of GDF-15 protein in cell lysates and

supernatants of selected cell lines analysed by Western blot (Figure 3B) and ELISA assay

11

(Figure 3C). Rates of proliferation of the 9 cell lines varied up to 3-fold, but failed to

correlate, neither in a positive or negative fashion, with levels of GDF-15 mRNA expression.

Since several studies have reported a strong p53 dependent induction of GDF-15 in a large

variety of cancers [4, 8-10], we additionally compared levels of GDF-15 expression with the

p53 status of the cell lines. Highest GDF-15 mRNA expression was found in 3 cell lines with

wt p53 status. However, p53 wt U87MG and mutant/p53 LN-18 cells showed lowest levels of

GDF-15 transcripts (Fig. 3A, Tab. 2) suggesting no significant correlation between the p53

status and the level of GDF-15 expression.

GDF-15 modulates proliferation of glioma cell lines

We next analysed, whether endogenous GDF-15 has an effect on the proliferation of

glioblastoma cells using a blocking antibody to GDF-15. As shown in Figure 4A, the antibody

(5µl/ml) significantly attenuated the mitogenic activity of 8 of the 9 human glioblastoma cell

lines. LN-18 cells showed no response to the antibody treatment in terms of proliferation.

Having shown that, with one exception, all glioblastoma cell lines employed in this study

responded to endogenous GDF-15, we next investigated their responsiveness to exogenously

applied GDF-15. As shown in Figure 4A, addition of recombinant GDF-15 (10ng/ml) to cell

cultures promoted proliferation of A172, D-247MG, LN-18, LN-319, but had no effect on

proliferation of LN-229, LN-308, LN-428, U373MG, and U373MG cells.

Taken together, these data suggest that endogenous GDF-15 is implicated in promoting

growth of most glioma cell lines studied, and that exogenous GDF-15 exerts growth

promoting effects on a more restricted set of glioma cell lines.

12

GDF-15 modulates invasiveness of distinct glioma cell lines

Invasiveness is a key feature of glioblastoma cells [28]. Figure 4B reveals that neutralization

of endogenous GDF-15 significantly suppressed invasiveness of LN-308 and U373MG

human glioma cells, promoted invasiveness of LN-428 cells, and did not affect invasiveness

of the other cell lines studied. Exogenous GDF-15 did not interfere with invasiveness of any

glioma cell line investigated.

Together, these data indicate that endogenous, but not exogenous GDF-15 affects

invasiveness of several but not all glioma cell lines studied.

GDF-15 does not affect apoptosis in glioma cell lines

Previous studies have identified GDF-15 as an important mediator of DNA damage in various

cancer cell lines, explaining its largely anti-tumorigenic and proapoptotic effects [4, 8, 14]. To

explore whether GDF-15 affects apoptosis in glioblastoma cell lines we quantified the number

of TUNEL positive cells in cultures previously treated with recombinant GDF-15 or GDF-15

antiserum (Table 3). Comparisons with control cultures showed no difference in numbers of

TUNEL positive cells suggesting that neither exogenous GDF-15 nor the inhibition of

endogenous GDF-15 have an impact on apoptotic cell death in the glioblastoma cell lines

investigated.

13

Discussion

The present study provides evidence for expression of GDF-15 in select human glioblastoma

tissues, glioblastoma primary cultures, and glioblastoma cell lines. While GDF-15

immunoreactivity was prominent in 3 of the 4 investigated secondary glioblastomas, only one

of nine primary glioblastomas showed staining for GDF-15. In support of the notion that

primary and secondary glioblastomas may differ with regard to their GDF-15 protein

expression, quantitative real-time RT-PCR carried out on cultured primary and secondary

glioblastoma cells revealed higher levels of GDF-15 transcripts in secondary as compared to

primary glioblastoma cells. With regard to secondary glioblastoma cell cultures we found an

overall high expression in all WHO grades, but no differences that could be correlated to the

degree of malignancy. This suggests that GDF-15 is unlikely to hold a key position in terms

of promoting progression of glioblastoma from a less to a more malignant phenotype

It is well established that the genetic pathways of primary and secondary glioblastomas

leading to glioma as the common phenotypic endpoint are different and show little overlap

[29]. Thus, higher expression levels of GDF-15 mRNA in secondary than primary

glioblastomas might be regulated by one or several of the known mutated, down- or

upregulated signalling pathways. In this context, p53 mutations were among the first genetic

alterations identified in glioma tumors. p53 mutations are less common in primary

glioblastoma (< 10%), while secondary glioblastomas have a high incidence of p53 mutations

(>65%). In the light of studies that have reported a strong p53 dependent induction of GDF-15

in a large variety of cancers [4, 8-10], we had expected to detect lower levels of GDF-15

expression in secondary glioblastoma because of increased probability for p53 mutations. One

explanation for this somewhat unexpected result might be a possible p53 independent

regulation of GDF-15 gene expression. p53 independent upregulation of GDF-15 has

previously been shown as a result of DNA damage, induced by genotoxic agents, irradiation

14

[8], or treatments with nonsteroidal anti-inflammatory drugs [4]. Thus, it is likely that GDF-

15 can be regulated through signals and pathways not involving p53.

Immunocytochemistry carried out on primary and secondary glioblastoma tissues revealed

clear differences in the localization of GDF-15 protein. In the primary glioblastoma GDF-15

immunoreactivity was exclusively localized in tumor cells, but absent from blood vessels and

perivascular areas. In contrast, we found prominent GDF-15 immunoreactivity in the ECM of

perivascular spaces of 2 anaplastic astrocytoma grade III (AAG III) and one secondary

glioblastoma grade IV (SGG IV), whereas tumor cells and microglia were GDF-15-negative.

Thus, it appears that GDF-15 of secondary glioblastomas is primarily located in the ECM.

Stromal ECM stores for GDF-15 in prostate tumors have previously been shown by Bauskin

and co-workers [30]. Interestingly, increased stromal stores of GDF-15 in prostate tumors

were associated with a significantly greater chance of remaining disease free. Whether

stromal storage of proGDF-15 in the ECM has a biological significance in secondary

glioblastomas remains to be elucidated.

To address putative roles of GDF-15 for glioblastoma cell properties, we investigated GDF-

15 expression and functions in a panel of 9 human glioblastoma cell lines with different p53

backgrounds. We found only weak GDF-15 mRNA expression in cell lines showing p53

mutations or deletions. Two of the four cell lines with wild type p53 status showed a higher

GDF-15 expression rate. This difference was further reflected in the amount of GDF-15

protein levels. Whether or not the increased expression depends on p53 remains to be

investigated.

Previous studies have identified GDF-15 as an important downstream mediator of DNA

damage signalling in various cancer cell lines and anti-tumorigenic and proapoptotic effects

of GDF-15 have been reported [4, 8, 16]. Moreover, recent reports demonstrated that

transfection of GDF-15 cDNA into human colorectal carcinoma and glioblastoma cells

15

decreased tumor growth in vivo and in vitro [4, 19]. In the present study we were unable to

demonstrate that recombinant GDF-15 activates apoptosis and/or suppresses proliferation of

glioblastoma cell lines. Our analyses rather revealed that exogenously added GDF-15 has no

effect or can stimulate proliferation of glioblastoma cell lines. More importantly, neutralizing

antibodies against GDF-15 attenuated the mitogenic activity of 8 (out of nine) cell lines

suggesting that blocking antibodies or Fab fragments of GDF-15 might be developed into a

therapeutic tool for treating glioblastoma. Together, the present and previous sets of data do

not allow to assign unequivocal promoting or inhibitory roles to GDF-15 in the regulation of

cell proliferation.

GDF-15 ambiguously affected glioma cell invasiveness in this study. Other studies carried out

with other types of cancer cells have shown that recombinant GDF-15 can promote

invasiveness of gastric cancer cells in a dose dependent manner [31]. Similarly, loss of cell

adhesion has been described in prostate cancer cells upon GDF-15 treatment [16]. Given the

diversity of effects exerted by GDF-15 in relation to invasiveness depending on cancer cell

type, it is conceivable that GDF-15, as other members of the TGF-ß superfamily, may

function in a crucially context-dependent manner

In conclusion, we have provided evidence for distinct levels and sites of expression of GDF-

15, a distant member of the TGF-ß superfamily, in primary and secondary glioblastoma. In

contrast to the documented proapoptotic and anti-tumorigenic activities of GDF-15 in several

cancer cell lines, our data suggest that GDF-15 does not decrease proliferation of

glioblastoma cell lines, while its effects on invasiveness are inconsistent. Mechanisms that

determine the distinct directions of the effects of GDF-15 may be sought in its context-

dependent actions with other growth factors.

16

Acknowledgements

This work was supported by grants of the Deutsche Forschungsgemeinschaft (STR 616/3-4,

KU 34/23-1) and the DFG Graduate College 791 at Heidelberg University. “Neural

Developmental and Degenerative Processes; Basic Research and Clinical Implications”.

17

References

[1] M.R. Bootcov, A.R. Bauskin, S.M. Valenzuela, et al., MIC-1, a novel macrophage inhibitory cytokine, is a

divergent member of the TGF-beta superfamily, Proc. Natl. Acad. Sci. USA 94 (1997) 11514-11519.

[2] V.M. Paralkar, A.L. Vail, W.A. Grasser, et al., Cloning and characterization of a novel member of the

transforming growth factor-beta/bone morphogenetic protein family, J. Biol. Chem. 273 (1998) 13760-

13767.

[3] R. Hromas, M. Hufford, J. Sutton, D. Xu, Y. L, L. Lu, PLAB, a novel placental bone morphogenetic protein,

Biochem. Biophy, Acta 1354 (1997) 40-44

[4] S.J. Baek, K.S. Kim, J.B. Nixon, L.C. Wilson, T.E. Eling, Cyclooxygenase inhibitors regulate the expression

of a TGF-beta superfamily member that has proapoptotic and antitumorigenic activities, Mol. Pharmacol.

59 (2001) 901-908.

[5] M. Böttner, C. Suter-Crazzolara, A. Schober, K. Unsicker, Expression of a novel member of the TGF-beta

superfamily, growth/differentiation factor-15/macrophage-inhibiting cytokine-1 (GDF-15/MIC-1) in adult

rat tissues, Cell Tissue Res. 297 (1999) 103-110.

[6] J. Strelau, A. Sullivan, M. Böttner, et al., Growth/differentiation factor-15/macrophage inhibitory cytokine-1

is a novel trophic factor for midbrain dopaminergic neurons in vivo, J. Neurosci. 20 (2000) 8597-8603.

[7] M. Böttner, K. Krieglstein, K. Unsicker, The transforming growth factor-betas: structure, signaling, and roles

in nervous system development and functions, J. Neurochem. 75 (2000) 2227-2240.

[8] P.-X. Li, J. Wong, A. Ayed, et al., Placental transforming growth factor- is a downstream mediator of the

growth arrest and apoptotic response of tumor cells to DNA damage and p53 overexpression, J. Biol.

Chem. 275 (2000) 20127-20135.

[9] K. Kannan, N. Amariglio, G. Rechavi, D. Givol, Profile of gene expression regulated by induced p53:

connection to the TGF- family, FEBS Lett. 470 (2000) 77-82.

[10] M. Tan, Y. Wang, K. Guan, Y. Sun, PTGF-, a type transforming growth factor (TGF-) superfamily

member, is a p53 target gene that inhibits tumor cell growth via TGF- signaling pathway, Proc. Natl.

Acad. Sci. USA 97 (2000) 109-114.

[11] J. Wong, P.X. Li, H.J. Klamut, A novel p53 transcriptional repressor element (p53TRE) and the

asymmetrical contribution of two p53 binding sites modulate the response of the placental transforming

growth factor-beta promoter to p53, J. Biol. Chem. 277 (2002) 26699-26707.

18

[12] F.G. Bottone Jr., S.J. Baek, J.B. Nixon, T.E. Eling, Diallyl disulfide (DADS) induces the antitumorigenic

NSAID-activated gene (NAG-1) by a p53-dependent mechanism in human colorectal HCT 116 cells, J.

Nutr. 132 (2002) 773-778.

[13] S.J. Baek, L.C. Wilson, C.-H. Lee, T.E. Eling, Dual function of nonsteroidal anti-inflammatory drugs

(NSAIDs): inhibition of cyclooxygenase and induction of NSAID-activated gene, J. Pharmacol. Exp.

Ther. 301 (2002) 1126-1131.

[14] P. Buckhaults, C. Rago, B. St Croix, et al., Secreted and cell surface genes expressed in benign and

malignant colorectal tumors, Cancer Res. 61 (2001) 6996-7001.

[15] R. Thomas, L.D. True, P.H. Lange, R.L.Vessella, Placental bone morphogenetic protein (PLAB) gene

expression in normal, pre-malignant and malignant human prostate: relation to tumor development and

progression, Int. J. Cancer 93 (2001) 47-52.

[16] T. Liu, A.R. Bauskin, J. Zaunders, et al., Macrophage inhibitory cytokine 1 reduces cell adhesion and

induces apoptosis in prostate cancer cells, Cancer Res. 63 (2003) 5034-5040.

[17] J.B. Welsh, L.M. Sapinoso, S.G. Kern, et al., Large-scale delineation of secreted protein biomarkers

overexpressed in cancer tissue and serum, Proc. Natl. Acad. Sci. USA 100 (2003) 3410-3415.

[18] D. Brown, R.L. Ward, P. Buckhaults, et al., MIC-1 serum level and genotype: associations with progress

and prognosis of colorectal carcinoma, Clin. Cancer Res. 9 (2003) 2642-2650.

[19] M. Albertoni, P.H. Shaw, M. Nozaki, et al., Anoxia induces macrophage inhibitory cytokine-1 (MIC-1) in

glioblastoma cells independently of p53 and HIF-1, Oncogene 21 (2002) 4212-4219.

[20] R. Graichen, D. Liu, Y. Sun, K.O. Lee, P.E. Lobie, Autocrine human growth hormone inhibits placental

transforming growth factor-beta gene transcription to prevent apoptosis and allow cell cycle progression

of human mammary carcinoma cells, J. Biol. Chem. 277 (2002) 26662-26672.

[21] M. Shim, T.E. Eling, Protein kinase C-dependent regulation of NAG-1/placental bone morphogenic

protein/MIC-1 expression in LNCaP prostate carcinoma cells, J. Biol. Chem. 280 (2005) 18636-18642.

[22] S. Karcher, H.H. Steiner, R. Ahmadi,et al., Different angiogenic phenotypes in primary and secondary

glioblastomas, Int. J. Cancer 118 (2006) 2182-2189.

[23] A. Schober, M. Böttner, J. Strelau, et al., Expression of growth differentiation factor-15/ macrophage

inhibitory cytokine-1 (GDF-15/MIC-1) in the perinatal, adult, and injured rat brain, J. Comp. Neurol. 439

(2001) 32-45.

19

[24] A.W. Henkel, S.C. Bieger, Quantification of proteins dissolved in an electrophoresis sample buffer, Anal.

Biochem. 223 (1994) 329-331.

[25] M. Gupta, A. Djalilvand, D.J. Brat. Clarifying the diffuse gliomas: an update on the morphologic features

and markers that discriminate oligodendroglioma from astrocytoma, Am J Clin Pathol. 124 (2005) 755-

768.

[26] C. Hartmann, W. Mueller, A.von Deimling. Pathology and molecular genetics of oligodendroglial tumors. J

Mol Med. 82 (2004) 638-655.

[27] H. Ohgaki, Genetic pathways to glioblastomas, Neuropathology 25 (2005) 1-7.

[28] T. Demuth, M.E. Berens, Molecular mechanisms of glioma cell migration and invasion, J. Neurooncol. 70

(2004) 217-228.

[29] P. Kleihues, W.K Cavenee, (Eds.) Pathology and genetics of tumours of the nervous system. IARC Press

Lyon, 2000, pp9-54.

[30] A.R. Bauskin, D.A. Brown, S. Junankar, et al., The propeptide mediates formation of stromal stores of

PROMIC-1: role in determining prostate cancer outcome, Cancer Res. 65 (2005) 2330-2336.

[31] D.H. Lee, Y. Yang, S.J. Lee, et al., Macrophage inhibitory cytokine-1 induces the invasiveness of gastric

cancer cells by up-regulating the urokinase-type plasminogen activator system, Cancer Res. 63 (2003)

4648-4655.

20

Figure Legends

Fig. 1A. Presence of GDF-15 in human primary and secondary glioblastoma tissue samples.

Left panel, peroxidase immunostaining of GDF-15 in a primary glioblastoma counterstained

with hematoxylin. Tumor cells are GDF-15 positive (T), whereas no staining was detectable

in the stroma of the tumor (S). Right panel, in the negative control (without 1st antibody), no

unspecific binding was observed. Fig. 1B. Fluorescence immunostainings of GDF-15, CD68

and GFAP in serial sections of an anaplastic astrocytoma grade III. Upper left panel

represents an overlay of GDF-15 and DAPI staining. Notably, GDF-15 is present in the

perivascular spaces of the stroma. Upper right panel: negative control using GDF-15 pre-

immune serum and DAPI staining. Bottom left panel: GFAP positive tumor cells are negative

for GDF-15. Bottom right panel: CD68 positive microglia do not show GDF-15

immunoreactivity.

Fig. 2. GDF-15 expression in primary cultures from human brain tumors of different origin

and malignancy grades was analyzed by quantitative RT-PCR. GDF-15 expression was

normalized against GAPDH and ß-Actin. PGG IV showed a significantly lower GDF-15

expression compared to SGG IV or OAG II (p<0.05). Comparison of PGG IV with OGG II or

OGG III revealed a significant difference of p<0.01. DAG II: diffuse astrocytomas grade II;

AAG III: anaplastic astrocytomas grade III; SGG IV: secondary glioblastomas grade IV; PGG

IV: primary glioblastomas grade IV: oligodendroglioblastomas grade II and III; OGG II and

III: oligoastrozytomas grade II and III; OAG II and III. Bars represent mean ± SD.

Fig. 3. No correlation exists between GDF-15 expression and rates of proliferation in

glioblastoma cell lines. (A) The upper figure shows GDF-15 expression levels in nine human

glioblastoma cell lines (quantitative RT-PCR), the lower figure shows proliferation rates of

21

the same cell lines, measured by BrdU incorporation. Bars represent ± SD. (B) Immunoblots

of cell lysates with purified GDF-15 antiserum confirm the highest expression of GDF-15

protein in D-247MG and A-172 cells. (C) GDF-15 protein concentrations per 10.000 cells in

supernatants of D-247MG, LN-308 and U87MG using an ELISA assay.

Fig. 4. Effects of GDF-15 anti-serum and recombinant human GDF-15 on the rates of

proliferation and invasive capacities of glioblastoma cell lines. (A) Proliferation assays

performed on ten glioblastoma cell lines show that the growth of most cell lines is inhibited

by GDF-15 anti-serum and slightly stimulated by recombinant GDF-15. (B) Invasion assays

indicate that recombinant GDF-15 does not affect invasiveness of glioblastoma cell lines.

However, GDF-15 anti-serum does significantly affect the invasive capacity of three of the

glioblastoma cell lines. * p<0.05; ** p<0.01; *** p<0.001. Bars represent ± SD.

1

Cancer Letters

EXPRESSION AND PUTATIVE FUNCTIONS OF GDF-15, A MEMBER

OF THE TGF-ß SUPERFAMILY, IN HUMAN GLIOMA AND

GLIOBLASTOMA CELL LINES

Jens Strelau 1, Corina Schmeer 1, Heike Peterziel 1, Tina Paech 1, Christel Herold-

Mende 2, Hans Steiner 3, Michael Weller 4 and Klaus Unsicker 1

1Department of Neuroanatomy and Interdisciplinary Center for Neurosciences (IZN)

University of Heidelberg, Germany2 Molecular Biology Laboratories, Department of Neurosurgery, University of Heidelberg,

Germany 3 Department of Neurosurgery, Nürnberg Süd, Nürnberg, Germany4 Department of Neurology, University Hospital Zurich, Switzerland.

Address for correspondence:

Dr. Jens Strelau

Neuroanatomy & Interdisciplinary Center for Neurosciences (IZN),

University of Heidelberg,

INF 307

D-69120 Heidelberg, Germany

Fax: 49-6221-54 56 04

e-mail: jens.strelau@urz.uni-heidelberg.de

* Manuscript with track changes

2

Abstract

Recent studies have demonstrated growth-inhibiting effects of Growth differentiation factor-

15 (GDF-15) on different cancer cell lines in vitro and on tumor growth in vivo. Here, we

present data concerning expression of GDF-15 in glioblastoma. We found low levels of GDF-

15 transcripts in primary glioblastoma. Thus, GDF-15 expression might be exploited as a

useful indicator for distinguishing primary from other glial derived tumors. In contrast to the

documented proapoptotic and anti-tumorigenic activities of GDF-15 in several cancer cell

lines, our data suggest that GDF-15 does not decrease proliferation of glioblastoma cell lines,

while its effects on invasiveness are not consistent.

Key words: GDF-15, TGF- superfamily, glioblastomas, tumor

3

Introduction

Growth differentiation factor-15 (GDF-15), also known as macrophage inhibitory cytokine-1

[1], prostate-derived factor [2], placental bone morphogenetic protein [3], and nonsteroidal

anti-inflammatory drug-activated protein-1 [4], is a divergent member of the TGF-ß

superfamily and widely expressed in epithelial tissues and in brain [5, 6, 7].

Most recently, several reports have provided evidence for functional links between GDF-15

and factors involved in tumorigenesis. The tumor suppressor gene product p53, for instance,

strongly induces GDF-15, indicating for the first time a functional connection between p53

and the TGF-ß superfamily [8-12]. Nonsteroidal anti-inflammatory drugs (NSAIDs) that

inhibit tumor development can also increase GDF-15 expression [13]. Other studies have

documented increased levels of GDF-15 in serum and in tumors of patients with metastatic

cancer [14-18]. Perhaps most excitingly, GDF-15 has been reported to exert proapoptotic and

anti-tumorigenic functions on human colorectal, prostate, breast, and mammary cancer cells

in vitro and on colon and glioblastoma tumors in vivo [4, 8, 19-21]. Together, these findings

strongly suggest a putative role of GDF-15 in tumor development and/or progression and

underscore the importance of further exploration of an involvement of GDF-15 in

tumorigenesis.

We report here the expression of GDF-15 in human glioblastoma tissues, glioblastoma

primary cultures, and glioblastoma cell lines. Our analyses reveal that GDF-15 may be

implicated in the regulation of proliferation and invasiveness of human glioblastoma cells.

4

Materials and Methods

Samples and Cell culture

Tissue specimens were obtained intraoperatively from central parts of the tumors. Primary

and secondary GBM were histopathologically classified according to the classification of the

WHO. Tumors were characterized as secondary GBM in patients with previous identified low

grade (WHO II and WHO III) glioblastomas. Primary cultures of glioma tissues (Tab. 1)

were established as described earlier [22]. The human glioma cell lines A172, D247MG, LN-

18, LN-229, LN-308, LN-319, and LN-428 (see Tab. 2) were maintained in DMEM with 4.5

g glucose/L supplemented with 10% fetal bovine serum (FBS), penicillin and streptomycin.

Cultures were maintained in a humidified 5% CO2 atmosphere at 37oC.

GDF-15 immunohistochemistry

Samples of primary and secondary glioblastomas were obtained from surgical cases in the

Department of Neurosurgery, University of Heidelberg, Germany. Snap frozen samples were

embedded in Tissue-Tek (Sakura, Ramsey, MI, USA) and directly cut into 16µm slices. For

immunohistochemistry, sections were fixed in 4% PFA for 1 hour. Non-specific peroxidase

activity was blocked by incubating sections in 1% hydrogen peroxide in PBS for 10 min at

room temperature (RT). Non-specific antibody binding was blocked using blocking buffer

(5% goat serum and 0.1% Triton X-100 in PBS) for 1 hour at RT. After washing, coverslips

were incubated with rabbit anti-mouse GDF-15 polyclonal antibodies (affinity-purified and

diluted 1:200) or with PBS overnight at 4oC. Slides were subsequently washed 3 times for 5

min in PBS and incubated 30 min at RT with biotinylated goat anti-rabbit IgG (Vectastain

Elite ABC Kit, Vector Laboratories, Burlingame, CA, USA). All subsequent steps were

performed according to the manufacturer’s instructions. Slides were counterstained with

hematoxylin, coverslipped, and photographed using a Zeiss Axioplan 2 equipped with

5

AxioVision 3.1 software. Immunofluorescence was performed as described [23]. Primary

antibodies used were affinity purified polyclonal anti-GDF-15, monoclonal anti-CD68 (BD

Biosciences, San Jose, CA, USA), and monoclonal anti-GFAP (Sigma, Chicago, IL, USA).

Secondary antibodies were anti-mouse IgG Cy2 and anti-rabbit IgG Cy3 (Jackson

ImmunoResearch, West Baltimore Pike West Grove, PA, USA)

Gel electrophoresis and immunoblot analyses

Protein extracts were prepared by homogenizing the cells in electrophoresis sample buffer,

and the protein content was determined using densitometry [24]. 25µg protein extract per lane

were loaded on SDS-polyacrylamide gels and transferred to nitrocellulose membranes

(Hybond ECL, Amersham Pharmacia, Göttingen, Germany) by electroblotting. The

membranes were incubated with purified polyclonal rabbit anti-rat GDF-15/MIC-1 antibody

for 16 h at 4 °C. Bound antibody was detected with a peroxidase-conjugated secondary

antibody and the ECL western blotting substrate system (Amersham Pharmacia) according to

the manufacturer´s manual. Samples of purified recombinant GDF-15 were visualized with

Coomassie blue and quantified by densitometry comparison [24] with defined concentrations

of protein standards.

ELISA

For the determination of secreted GDF-15 from the glioma cell lines, 25.000 cells/well were

seeded in 48 well plates. The cell culture supernatants were collected after 24 hours and the

amount of GDF-15 determined by sandwich ELISA using a monoclonal anti-human GDF-15

capture antibody (R&D Systems, MAB957) and a biotinylated goat anti-human GDF-15

detection antibody (R&D Systems, BAF940) according to the manufacturer’s protocol.

Recombinant human GDF-15 (R&D Systems) diluted in medium (15 pg/ml to 250 pg/ml)

6

was used as a standard for the determination of the total amount of GDF-15 secreted by the

cells. The number of cells per well after 14 hours was determined by direct counting. The

experiment was performed three times with every cell line tested in triplicate per experiment.

Total RNA extraction and cDNA synthesis

Samples of tumor tissue (100 – 200 mg each) and cultured glioma cells were lysed in

peqGOLD TriFast reagent (Peqlab, Erlangen, Germany), and phenol-chloroform extraction of

total RNA was performed. Prior to reverse transcription RNA samples were treated with RQ1

DNase (Promega). cDNA synthesis was then performed using M-MLV Reverse Transcriptase

(Promega). Additionally, total RNA was extracted from primary (low-passage) cultures

established from glioma tissue of different malignancy grades.

Real-time quantitative RT-PCR analysis

Relative quantitation of human GDF-15 and GAPDH transcripts in cDNA samples was

performed on an ABI PRISM 7000 Sequence Detection System using the corresponding

TaqMan Assays-on-Demand Gene Expression Products (Applied Biosystems) and following

the manufacturer’s protocol. Each sample was analyzed in triplicate. GAPDH expression was

considered as an internal control to which GDF-15 expression was normalized.

Cell proliferation assays

Cells were seeded onto 96-well plates at a density of 104 (for U373MG: 5 x 103) cells/well.

On the next day, cultures were treated with rabbit anti-mouse GDF-15 polyclonal serum (5

L/ml), pre-immune serum (5 l/ml) as a control, or human recombinant GDF-15 (20 ng/ml)

(R&D). 24 hours later, cell proliferation was measured using 5-Bromo-2`-deoxy-uridine

7

labeling and the Detection Kit III (Roche Applied Science), following the manufacturer’s

instructions.

In vitro cell invasion assays

The ability of cells to migrate through Matrigel-coated filters was determined using the 24-

well plate BD BioCoat Matrigel Invasion Chamber (BD Biosciences, Bedford, MA, USA).

Briefly, cells were seeded at a density of 2.5x104 cells in 500 L DMEM onto Matrigel

inserts. Rabbit anti-mouse GDF-15 polyclonal serum (5 l/m), pre-immune serum (5 l/ml, as

a control), or human recombinant GDF-15 (20 ng/ml) were added to the inserts. Wells of the

24-well plates contained DMEM supplemented with 5% FBS as a chemoattractant. After

incubation for 24 h at 37oC in 5% CO2 the cells that had not invaded the filters were removed

with cotton swabs, and cells that had migrated to the lower surface of the filter were fixed in

methanol. Cells were then stained with Toluidine Blue (Merck, Darmstadt, Germany) and

counted in 32 fields (x200) per filter, covering the periphery and the center of the membrane.

All experiments were performed in triplicate.

TUNEL

Cells were washed and fixed with 4% PFA, permeabilized with 0.2% Triton X-100, and

processed for TUNEL (Promega). 5 µg/ml DAPI was added to stain cell nuclei.

Photomicrographs from 4-6 different fields in each coverslip were captured under green

channels and merged using adobe photoshop 5.5. Typically, about 1000 cells were analyzed

for the number of TUNEL-positive glioblastoma cells.

8

Statistical analysis

Data were expressed as means ± SEM. All experiments were performed in triplicates or

duplicates and repeated at least three times. Statistical analyses among groups were performed

using the t test (Origin 6.1). P-values are as follows: * P < 0.05, ** P < 0.01, and *** P <

0.001.

9

Results

GDF-15 immunoreactivity in de novo glioblastoma, high grade astrocytoma and

secondary glioblastoma

We first investigated the possible presence of GDF-15 protein in surgically removed human

primary and secondary glioblastomas using immunohistochemistry. Only one out of the nine

investigated primary glioblastomas grade IV showed moderate GDF-15 immunoreactivity (ir)

(Figure 1A; left panel). GDF-15 staining was localized to tumor cells, but absent from blood

vessels. The right panel shows the control experiment where the GDF-15 antibody was

omitted. In contrast to the largely GDF-15 negative primary glioblastomas we found

prominent staining for GDF-15 ir in 2 anaplastic astrocytomas grade III (AAG III) and one

secondary glioblastoma. The upper left panel of Figure 1B exemplifies GDF-15 ir in one of

the AAG III. Notably, the stroma and perivascular spaces of the AAG III were

immunoreactive for GDF-15. An oligodendroglioma grade III (OG III) was negative for

GDF-15 ir. In samples of tumor tissues, where GDF-15 ir was detectable, there was no

costaining of GDF-15 with glial fibrillary acidic protein (GFAP; Figure 1B; bottom left

panel). Similarly, GDF-15 immunoreactive cells in AAG III (Figure 1B; bottom right panel)

and secondary glioblastomas did not show co-labeling for the microglial marker CD68.

Together, these data suggest the presence of GDF-15 protein in GFAP- and CD68-negative

cells, preferentially in AAG III and secondary glioblastoma.

Lowest expression of GDF-15 in cultures of primary glioblastomas

Diffuse astrocytomas grade II (DAG II), anaplastic astrocytomas grade III (AAG III), and

secondary glioblastomas grade IV (SGG IV) represent distinct stages in the evolution of

secondary glioblastomas, whereas primary glioblastomas grade IV (PGG IV) develop along a

different genetic pathway [25]. Other brain tumours distinguishable from those mentioned

10

above due to their distinct genetical alteration, mainly characterized by a combined loss of

chromosomes 1p and 19q, are oligodendroglioblastomas and oligoastrocytomas grade II and

III (OGG II, OGG III; OAG II, OAG III) [26,27]. Figure 2 shows the results of quantitative

real-time PCR determinations of GDF-15 mRNA expressed in cultures derived from 10 DAG

II, 10 AAG III, 10 SGG IV, 10 PGG IV, 5 OGG II, 5 OGG III, 5 OAG II, and 5 OAG III.

High and largely comparable expression levels were found in DAG II, AAG III, and SGG IV

(Figure 2). In contrast, PGG IV showed significantly lower levels of GDF-15 expression. This

difference was even more pronounced when GDF-15 expression in PGG IV was compared to

the average expression levels of cell cultures derived from oligodendroglioblastomas grade II

(OGG II) and anaplastic oligodendroglioblastomas grade III (OGG III). The most frequent

type of mixed glioblastomas, OAG II, also showed a significantly higher GDF-15 level

compared to PGG IV. Thus, differences in GDF-15 expression might be exploited as a useful

indicator for distinguishing primary from other glial derived tumors.

GDF-15 expression does not correlate with spontaneous proliferation or the p53 status

of glioma cell lines

In an attempt to elucidate putative functions of GDF-15 on glioblastoma growth we

investigated a panel of 9 human glioma cell lines (LN-319, U87MG, A172, D-247MG, LN-

229, LN-308, LN-428, LN-18 and U373MG). We first compared the expression levels of

endogenous GDF-15 mRNA in relation to the spontaneous proliferation rate for each human

cell line. Figure 3A shows the wide range of mRNA expression levels in these cell lines, from

very high rates in A172 and D-247MG cells to up to 15-fold lower levels in LN-18 and

U87MG cells. Similar differences were found in terms of GDF-15 protein in cell lysates and

supernatants of selected cell lines analysed by Western blot (Figure 3B) and ELISA assay

11

(Figure 3C). Rates of proliferation of the 9 cell lines varied up to 3-fold, but failed to

correlate, neither in a positive or negative fashion, with levels of GDF-15 mRNA expression.

Since several studies have reported a strong p53 dependent induction of GDF-15 in a large

variety of cancers [4, 8-10], we additionally compared levels of GDF-15 expression with the

p53 status of the cell lines. Highest GDF-15 mRNA expression was found in 3 cell lines with

wt p53 status. However, p53 wt U87MG and mutant/p53 LN-18 cells showed lowest levels of

GDF-15 transcripts (Fig. 3A, Tab. 2) suggesting no significant correlation between the p53

status and the level of GDF-15 expression.

GDF-15 modulates proliferation of glioma cell lines

(deleted text)

We next analysed, whether endogenous GDF-15 has an effect on the proliferation of

glioblastoma cells using a blocking antibody to GDF-15. As shown in Figure 4A, the antibody

(5µl/ml) significantly attenuated the mitogenic activity of 8 of the 9 human glioblastoma cell

lines. LN-18 cells showed no response to the antibody treatment in terms of proliferation.

Having shown that, with one exception, all glioblastoma cell lines employed in this study

responded to endogenous GDF-15, we next investigated their responsiveness to exogenously

applied GDF-15. As shown in Figure 4A, addition of recombinant GDF-15 (10ng/ml) to cell

cultures promoted proliferation of A172, D-247MG, LN-18, LN-319, but had no effect on

proliferation of LN-229, LN-308, LN-428, U373MG, and U373MG cells.

Taken together, these data suggest that endogenous GDF-15 is implicated in promoting

growth of most glioma cell lines studied, and that exogenous GDF-15 exerts growth

promoting effects on a more restricted set of glioma cell lines.

12

GDF-15 modulates invasiveness of distinct glioma cell lines

Invasiveness is a key feature of glioblastoma cells [28]. Figure 4B reveals that neutralization

of endogenous GDF-15 significantly suppressed invasiveness of LN-308 and U373MG

human glioma cells, promoted invasiveness of LN-428 cells, and did not affect invasiveness

of the other cell lines studied. Exogenous GDF-15 did not interfere with invasiveness of any

glioma cell line investigated.

Together, these data indicate that endogenous, but not exogenous GDF-15 affects

invasiveness of several but not all glioma cell lines studied.

GDF-15 does not affect apoptosis in glioma cell lines

Previous studies have identified GDF-15 as an important mediator of DNA damage in various

cancer cell lines, explaining its largely anti-tumorigenic and proapoptotic effects [4, 8, 14]. To

explore whether GDF-15 affects apoptosis in glioblastoma cell lines we quantified the number

of TUNEL positive cells in cultures previously treated with recombinant GDF-15 or GDF-15

antiserum (Table 3). Comparisons with control cultures showed no difference in numbers of

TUNEL positive cells suggesting that neither exogenous GDF-15 nor the inhibition of

endogenous GDF-15 have an impact on apoptotic cell death in the glioblastoma cell lines

investigated.

13

Discussion

The present study provides evidence for expression of GDF-15 in select human glioblastoma

tissues, glioblastoma primary cultures, and glioblastoma cell lines. While GDF-15

immunoreactivity was prominent in 3 of the 4 investigated secondary glioblastomas, only one

of nine primary glioblastomas showed staining for GDF-15. In support of the notion that

primary and secondary glioblastomas may differ with regard to their GDF-15 protein

expression, quantitative real-time RT-PCR carried out on cultured primary and secondary

glioblastoma cells revealed higher levels of GDF-15 transcripts in secondary as compared to

primary glioblastoma cells. With regard to secondary glioblastoma cell cultures we found an

overall high expression in all WHO grades, but no differences that could be correlated to the

degree of malignancy. This suggests that GDF-15 is unlikely to hold a key position in terms

of promoting progression of glioblastoma from a less to a more malignant phenotype

It is well established that the genetic pathways of primary and secondary glioblastomas

leading to glioma as the common phenotypic endpoint are different and show little overlap

[29]. Thus, higher expression levels of GDF-15 mRNA in secondary than primary

glioblastomas might be regulated by one or several of the known mutated, down- or

upregulated signalling pathways. In this context, p53 mutations were among the first genetic

alterations identified in glioma tumors. p53 mutations are less common in primary

glioblastoma (< 10%), while secondary glioblastomas have a high incidence of p53 mutations

(>65%). In the light of studies that have reported a strong p53 dependent induction of GDF-15

in a large variety of cancers [4, 8-10], we had expected to detect lower levels of GDF-15

expression in secondary glioblastoma because of increased probability for p53 mutations. One

explanation for this somewhat unexpected result might be a possible p53 independent

regulation of GDF-15 gene expression. p53 independent upregulation of GDF-15 has

previously been shown as a result of DNA damage, induced by genotoxic agents, irradiation

14

[8], or treatments with nonsteroidal anti-inflammatory drugs [4]. (deleted text) Thus, it is

likely that GDF-15 can be regulated through signals and pathways not involving p53.

Immunocytochemistry carried out on primary and secondary glioblastoma tissues revealed

clear differences in the localization of GDF-15 protein. In the primary glioblastoma GDF-15

immunoreactivity was exclusively localized in tumor cells, but absent from blood vessels and

perivascular areas. In contrast, we found prominent GDF-15 immunoreactivity in the ECM of

perivascular spaces of 2 anaplastic astrocytoma grade III (AAG III) and one secondary

glioblastoma grade IV (SGG IV), whereas tumor cells and microglia were GDF-15-negative.

Thus, it appears that GDF-15 of secondary glioblastomas is primarily located in the ECM.

Stromal ECM stores for GDF-15 in prostate tumors have previously been shown by Bauskin

and co-workers [30]. Interestingly, increased stromal stores of GDF-15 in prostate tumors

were associated with a significantly greater chance of remaining disease free. Whether

stromal storage of proGDF-15 in the ECM has a biological significance in secondary

glioblastomas remains to be elucidated.

To address putative roles of GDF-15 for glioblastoma cell properties, we investigated GDF-

15 expression and functions in a panel of 9 human glioblastoma cell lines with different p53

backgrounds. We found only weak GDF-15 mRNA expression in cell lines showing p53

mutations or deletions. Two of the four cell lines with wild type p53 status showed a higher

GDF-15 expression rate. This difference was further reflected in the amount of GDF-15

protein levels. Whether or not the increased expression depends on p53 remains to be

investigated.

Previous studies have identified GDF-15 as an important downstream mediator of DNA

damage signalling in various cancer cell lines and anti-tumorigenic and proapoptotic effects

of GDF-15 have been reported [4, 8, 16]. Moreover, recent reports demonstrated that

transfection of GDF-15 cDNA into human colorectal carcinoma and glioblastoma cells

15

decreased tumor growth in vivo and in vitro [4, 19]. In the present study we were unable to

demonstrate that recombinant GDF-15 activates apoptosis and/or suppresses proliferation of

glioblastoma cell lines. Our analyses rather revealed that exogenously added GDF-15 has no

effect or can stimulate proliferation of glioblastoma cell lines. More importantly, neutralizing

antibodies against GDF-15 attenuated the mitogenic activity of 8 (out of nine) cell lines

suggesting that blocking antibodies or Fab fragments of GDF-15 might be developed into a

therapeutic tool for treating glioblastoma. Together, the present and previous sets of data do

not allow to assign unequivocal promoting or inhibitory roles to GDF-15 in the regulation of

cell proliferation.

GDF-15 ambiguously affected glioma cell invasiveness in this study. Other studies carried out

with other types of cancer cells have shown that recombinant GDF-15 can promote

invasiveness of gastric cancer cells in a dose dependent manner [31]. Similarly, loss of cell

adhesion has been described in prostate cancer cells upon GDF-15 treatment [16]. Given the

diversity of effects exerted by GDF-15 in relation to invasiveness depending on cancer cell

type, it is conceivable that GDF-15, as other members of the TGF-ß superfamily, may

function in a crucially context-dependent manner

In conclusion, we have provided evidence for distinct levels and sites of expression of GDF-

15, a distant member of the TGF-ß superfamily, in primary and secondary glioblastoma. In

contrast to the documented proapoptotic and anti-tumorigenic activities of GDF-15 in several

cancer cell lines, our data suggest that GDF-15 does not decrease proliferation of

glioblastoma cell lines, while its effects on invasiveness are inconsistent. Mechanisms that

determine the distinct directions of the effects of GDF-15 may be sought in its context-

dependent actions with other growth factors.

16

Acknowledgements

This work was supported by grants of the Deutsche Forschungsgemeinschaft (STR 616/3-4,

KU 34/23-1) and the DFG Graduate College 791 at Heidelberg University. “Neural

Developmental and Degenerative Processes; Basic Research and Clinical Implications”.

17

References

[1] M.R. Bootcov, A.R. Bauskin, S.M. Valenzuela, et al., MIC-1, a novel macrophage inhibitory cytokine, is a

divergent member of the TGF-beta superfamily, Proc. Natl. Acad. Sci. USA 94 (1997) 11514-11519.

[2] V.M. Paralkar, A.L. Vail, W.A. Grasser, et al., Cloning and characterization of a novel member of the

transforming growth factor-beta/bone morphogenetic protein family, J. Biol. Chem. 273 (1998) 13760-

13767.

[3] R. Hromas, M. Hufford, J. Sutton, D. Xu, Y. L, L. Lu, PLAB, a novel placental bone morphogenetic protein,

Biochem. Biophy, Acta 1354 (1997) 40-44

[4] S.J. Baek, K.S. Kim, J.B. Nixon, L.C. Wilson, T.E. Eling, Cyclooxygenase inhibitors regulate the expression

of a TGF-beta superfamily member that has proapoptotic and antitumorigenic activities, Mol. Pharmacol.

59 (2001) 901-908.

[5] M. Böttner, C. Suter-Crazzolara, A. Schober, K. Unsicker, Expression of a novel member of the TGF-beta

superfamily, growth/differentiation factor-15/macrophage-inhibiting cytokine-1 (GDF-15/MIC-1) in adult

rat tissues, Cell Tissue Res. 297 (1999) 103-110.

[6] J. Strelau, A. Sullivan, M. Böttner, et al., Growth/differentiation factor-15/macrophage inhibitory cytokine-1

is a novel trophic factor for midbrain dopaminergic neurons in vivo, J. Neurosci. 20 (2000) 8597-8603.

[7] M. Böttner, K. Krieglstein, K. Unsicker, The transforming growth factor-betas: structure, signaling, and roles

in nervous system development and functions, J. Neurochem. 75 (2000) 2227-2240.

[8] P.-X. Li, J. Wong, A. Ayed, et al., Placental transforming growth factor- is a downstream mediator of the

growth arrest and apoptotic response of tumor cells to DNA damage and p53 overexpression, J. Biol.

Chem. 275 (2000) 20127-20135.

[9] K. Kannan, N. Amariglio, G. Rechavi, D. Givol, Profile of gene expression regulated by induced p53:

connection to the TGF- family, FEBS Lett. 470 (2000) 77-82.

[10] M. Tan, Y. Wang, K. Guan, Y. Sun, PTGF-, a type transforming growth factor (TGF-) superfamily

member, is a p53 target gene that inhibits tumor cell growth via TGF- signaling pathway, Proc. Natl.

Acad. Sci. USA 97 (2000) 109-114.

[11] J. Wong, P.X. Li, H.J. Klamut, A novel p53 transcriptional repressor element (p53TRE) and the

asymmetrical contribution of two p53 binding sites modulate the response of the placental transforming

growth factor-beta promoter to p53, J. Biol. Chem. 277 (2002) 26699-26707.

18

[12] F.G. Bottone Jr., S.J. Baek, J.B. Nixon, T.E. Eling, Diallyl disulfide (DADS) induces the antitumorigenic

NSAID-activated gene (NAG-1) by a p53-dependent mechanism in human colorectal HCT 116 cells, J.

Nutr. 132 (2002) 773-778.

[13] S.J. Baek, L.C. Wilson, C.-H. Lee, T.E. Eling, Dual function of nonsteroidal anti-inflammatory drugs

(NSAIDs): inhibition of cyclooxygenase and induction of NSAID-activated gene, J. Pharmacol. Exp.

Ther. 301 (2002) 1126-1131.

[14] P. Buckhaults, C. Rago, B. St Croix, et al., Secreted and cell surface genes expressed in benign and

malignant colorectal tumors, Cancer Res. 61 (2001) 6996-7001.

[15] R. Thomas, L.D. True, P.H. Lange, R.L.Vessella, Placental bone morphogenetic protein (PLAB) gene

expression in normal, pre-malignant and malignant human prostate: relation to tumor development and

progression, Int. J. Cancer 93 (2001) 47-52.

[16] T. Liu, A.R. Bauskin, J. Zaunders, et al., Macrophage inhibitory cytokine 1 reduces cell adhesion and

induces apoptosis in prostate cancer cells, Cancer Res. 63 (2003) 5034-5040.

[17] J.B. Welsh, L.M. Sapinoso, S.G. Kern, et al., Large-scale delineation of secreted protein biomarkers

overexpressed in cancer tissue and serum, Proc. Natl. Acad. Sci. USA 100 (2003) 3410-3415.

[18] D. Brown, R.L. Ward, P. Buckhaults, et al., MIC-1 serum level and genotype: associations with progress

and prognosis of colorectal carcinoma, Clin. Cancer Res. 9 (2003) 2642-2650.

[19] M. Albertoni, P.H. Shaw, M. Nozaki, et al., Anoxia induces macrophage inhibitory cytokine-1 (MIC-1) in

glioblastoma cells independently of p53 and HIF-1, Oncogene 21 (2002) 4212-4219.

[20] R. Graichen, D. Liu, Y. Sun, K.O. Lee, P.E. Lobie, Autocrine human growth hormone inhibits placental

transforming growth factor-beta gene transcription to prevent apoptosis and allow cell cycle progression

of human mammary carcinoma cells, J. Biol. Chem. 277 (2002) 26662-26672.

[21] M. Shim, T.E. Eling, Protein kinase C-dependent regulation of NAG-1/placental bone morphogenic

protein/MIC-1 expression in LNCaP prostate carcinoma cells, J. Biol. Chem. 280 (2005) 18636-18642.

[22] S. Karcher, H.H. Steiner, R. Ahmadi,et al., Different angiogenic phenotypes in primary and secondary

glioblastomas, Int. J. Cancer 118 (2006) 2182-2189.

[23] A. Schober, M. Böttner, J. Strelau, et al., Expression of growth differentiation factor-15/ macrophage

inhibitory cytokine-1 (GDF-15/MIC-1) in the perinatal, adult, and injured rat brain, J. Comp. Neurol. 439

(2001) 32-45.

19

[24] A.W. Henkel, S.C. Bieger, Quantification of proteins dissolved in an electrophoresis sample buffer, Anal.

Biochem. 223 (1994) 329-331.

[25] M. Gupta, A. Djalilvand, D.J. Brat. Clarifying the diffuse gliomas: an update on the morphologic features

and markers that discriminate oligodendroglioma from astrocytoma, Am J Clin Pathol. 124 (2005) 755-

768.

[26] C. Hartmann, W. Mueller, A.von Deimling. Pathology and molecular genetics of oligodendroglial tumors. J

Mol Med. 82 (2004) 638-655.

[27] H. Ohgaki, Genetic pathways to glioblastomas, Neuropathology 25 (2005) 1-7.

[28] T. Demuth, M.E. Berens, Molecular mechanisms of glioma cell migration and invasion, J. Neurooncol. 70

(2004) 217-228.

[29] P. Kleihues, W.K Cavenee, (Eds.) Pathology and genetics of tumours of the nervous system. IARC Press

Lyon, 2000, pp9-54.

[30] A.R. Bauskin, D.A. Brown, S. Junankar, et al., The propeptide mediates formation of stromal stores of

PROMIC-1: role in determining prostate cancer outcome, Cancer Res. 65 (2005) 2330-2336.

[31] D.H. Lee, Y. Yang, S.J. Lee, et al., Macrophage inhibitory cytokine-1 induces the invasiveness of gastric

cancer cells by up-regulating the urokinase-type plasminogen activator system, Cancer Res. 63 (2003)

4648-4655.

20

Figure Legends

Fig. 1A. Presence of GDF-15 in human primary and secondary glioblastoma tissue samples.

Left panel, peroxidase immunostaining of GDF-15 in a primary glioblastoma counterstained

with hematoxylin. Tumor cells are GDF-15 positive (T), whereas no staining was detectable

in the stroma of the tumor (S). Right panel, in the negative control (without 1st antibody), no

unspecific binding was observed. Fig. 1B. Fluorescence immunostainings of GDF-15, CD68

and GFAP in serial sections of an anaplastic astrocytoma grade III. Upper left panel

represents an overlay of GDF-15 and DAPI staining. Notably, GDF-15 is present in the

perivascular spaces of the stroma. Upper right panel: negative control using GDF-15 pre-

immune serum and DAPI staining. Bottom left panel: GFAP positive tumor cells are negative

for GDF-15. Bottom right panel: CD68 positive microglia do not show GDF-15

immunoreactivity.

Fig. 2. GDF-15 expression in primary cultures from human brain tumors of different origin

and malignancy grades was analyzed by quantitative RT-PCR. GDF-15 expression was

normalized against GAPDH and ß-Actin. PGG IV showed a significantly lower GDF-15

expression compared to SGG IV or OAG II (p<0.05). Comparison of PGG IV with OGG II or

OGG III revealed a significant difference of p<0.01. DAG II: diffuse astrocytomas grade II;

AAG III: anaplastic astrocytomas grade III; SGG IV: secondary glioblastomas grade IV; PGG

IV: primary glioblastomas grade IV: oligodendroglioblastomas grade II and III; OGG II and

III: oligoastrozytomas grade II and III; OAG II and III. Bars represent mean ± SD.

Fig. 3. No correlation exists between GDF-15 expression and rates of proliferation in

glioblastoma cell lines. (A) The upper figure shows GDF-15 expression levels in nine human

glioblastoma cell lines (quantitative RT-PCR), the lower figure shows proliferation rates of

21

the same cell lines, measured by BrdU incorporation. Bars represent ± SD. (B) Immunoblots

of cell lysates with purified GDF-15 antiserum confirm the highest expression of GDF-15

protein in D-247MG and A-172 cells. (C) GDF-15 protein concentrations per 10.000 cells in

supernatants of D-247MG, LN-308 and U87MG using an ELISA assay.

Fig. 4. Effects of GDF-15 anti-serum and recombinant human GDF-15 on the rates of

proliferation and invasive capacities of glioblastoma cell lines. (A) Proliferation assays

performed on ten glioblastoma cell lines show that the growth of most cell lines is inhibited

by GDF-15 anti-serum and slightly stimulated by recombinant GDF-15. (B) Invasion assays

indicate that recombinant GDF-15 does not affect invasiveness of glioblastoma cell lines.

However, GDF-15 anti-serum does significantly affect the invasive capacity of three of the

glioblastoma cell lines. * p<0.05; ** p<0.01; *** p<0.001. Bars represent ± SD.

Dear Editor:

thank you very much for sending us the constructive criticisms of two reviewers on our manuscript. We have dealt with their remarks in every respect (see below) and now submit a carefully revised version of the paper.

Response to Reviewers:

Reviewer 1

1. First this subtitle "GDF-15 is differentially expressed in cultures of primary and secondary glioblastoma" is misleading.We changed the subtitle into “Low expression of GDF-15 in cultures of primary glioblastomas” and present in the subsequent chapter data on diffuse astrocytomas grade II, anaplastic astrocytomas grade III, secondary glioblastomas grade IV, primary glioblastomas grade IV, oligodendroglioblastomas grade II and III, and oligoastrocytomas grade II and III.We also changed the first subtitle of the Results part into “GDF-15 immunoreactivity in de novo glioblastoma, high grade astrocytoma and secondary glioblastoma” and subdivided figure 1 into A and B for the sake of clarity of the immunohistochemical results.

2. No details are given about how these cultures were derived, maintained or what passage numbers were used in the present studies.Details are now provided in the Materials and Methods section on how the cultures were derived. Furthermore, data on GDF-15 expression in oligodendrogliomas grade II and III and oligoastrocytomas grade II and III were added to figure 2 and Results. We also incorporated a table comprising tumor type, passage numbers, age and sex of each specimen (Tab.1).

4. Did the authors conduct similar studies on the primary cell lines used in Figure 2?Figure 2 shows real-time PCR results with cDNA of 59 different primary cell cultures with low passage numbers provided by the Molecular Biology Laboratories, Department of Neurosurgery, University of Heidelberg. Since cell cultures with similarly low passage numbers were not available we did not perform invasion-, proliferation-, and TUNEL-assays on primary cell lines.

5. How do the authors explain the differences in responsiveness of the different cell lines?Our studies show that the GDF-15 antiserum decreases proliferation in 8 of 9 cell cultures. Exogenous GDF-15 increased cell division in 4 cell lines, while the remaining 5 cell lines did not respond. Several possibilities may be conceived to explain this result. First, cell lines unresponsive to exogenous GDF-15, but responsive to the anti-GDF-15 might already be maximally stimulated by the endogenous GDF-15. Moreover, non-responsive cell lines might require, in addition to GDF-15, a synergistically acting factor, which was not available at the required amounts. Alternatively, GDF-15 receptors in non-responsive cell lines might already be maximally saturated with endogenous GDF-15. A similar explanation might account for the lack of response to exogenous GDF-15 in the invasion experiments. It should be noted, however, that both U87MG and, to a lesser extent, LN-428 and LN-229, showed a tendency to respond to GDF-15.

It has been shown that, dependent on tumor type and origin (for review see Eling at al., 2006; Bauskin et al., 2006), GDF-15 can exhibit either tumorigenic or anti-tumorigenic activity. The cell lines investigated in our studies are derived from tumors of different stages and origin,and it is therefore conceivable that GDF-15 may regulate in either a positive or negative

* Revision notes

manner, depending on molecular and cellular contexts. Antagonistic effects have also been shown for TGF and its effects on different glioma cell lines (see also Jennings et al., 1991).

Although previous studies suggested that GDF-15 can regulate invasiveness in cancer cell lines, results of these former studies were based on a single cell line, without comparing the results with other cell lines derived from the same cancer types. It has been shown, for example, that GDF-15 significantly increases invasiveness of SNU-216, a human gastric cancer cell line (Lee et al., 2003). Evidently, this does not prove that all available gastric cancer cell lines would respond in the same way. Our study compares GDF-15 dependent invasion in 8 different cancer cell lines, which may differ in a variety of subtle respects.

6. Lastly, no effect on apoptosis was noted. How do the authors explain this result, which is different from previously published studies?Most in vitro studies suggest GDF-15- induces apoptosis in cancer cell lines. However, data from the literature provide high expression of GDF-15 in many tumors, a finding that is hard to be reconciled with the proposed tumor suppressor activity of GDF-15. Previous studies from our laboratory clearly demonstrate that GDF-15 can protect cerebellar granule cells from apoptotic cell death in vitro (Subramaniam et al., 2003), i.e. GDF-15 may function as an anti-apoptotic factor. To our knowledge, none of the previously published studies has shown pro-apoptotic effects of GDF-15 in glioma cultures.

7. The conclusions of this study are overstated.Conclusions have been changed

Reviewer 2

1. How are primary and secondary GBM differentiated (clinically or molecular). Primary and secondary GBM were differentiated as described earlier [22] In brief, tumors were histopathologically classified according to the classification of the WHO. Tumors were characterized as secondary GBM in patients with previous identified low grade (WHO II and WHO III) glioblastomas in accordance with Kleihues.

2. Why was no fresh material used for qPCR?We have now included fresh material provided by the Molecular Biology Laboratories, Department of Neurosurgery, University of Heidelberg. We got 9 frozen tissue samples ofprimary GBM and used the samples for qPCR and immunohistochemistry. The results of the qPCR with fresh primary GBM confirmed data obtained by IHC (low GDF-15 expression in 8 out of 9 tumors, (see also the figure on the last page of our response). Three of the primary cultures used for qPCR correspond to fresh tissue samples (Tab.1). Since one secondary GBM, two anaplastic astrocytoma and one oligodendroblastoma were provided as frozen sections on coverslips we could not perform qPCR and used the material directly for immunohistochemistry. However, primary cells derived from these tumors revealed high amounts of GDF-15 mRNA (Tab.1), although we were not able to show any ir in the oligodendroglioma grade III.

3. What is a low-passage culture and how was the immunohisto / qPCR correlation?

We added a list with tumor number, passage number, sex and age of the total material used in this study (Tab. 1).

4. Table 2 should be presented earlier in the results section.The correlation of p53 and GDF-15 is now mentioned in the Results section.

5. Discussion should be shortened.We now shortened the discussion.

This figure shows quantitative PCR of 9 primary tumors

Literature:

Asne R. Bauskin et al., Role of Macrophage Inhibitory Cytokine-1 in Tumorigenesis and Diagnosis of Cancer. Cancer Res. 2006; 66: (10). May 15, 4983-4986

Thomas E. Eling et al., NSAID Activated Gene (NAG-1), a Modulator of Tumorigenesis. J Biochem. and Mol. Biol. 2006; 39: (6). Nov. 649-655

Mark T. Jennings et al., TGF-ß1 and TGF-ß2 are potential growth regulators for low-grade and malignant gliomas in vitro: Evidence in support of an autocrine hypothesis. Int. J. Cancer 1991; 49, 129-139)

T

S

GFAP CD68

DAPI

Fig.1

GDF-15 + DAPI Control + DAPI

T

S

ControlGDF-15200µm

100µm

A

B

Figure(1)

0,0000

0,0200

0,0400

0,0600

0,0800

0,1000

0,1200

0,1400

DAG II AAG III SGG IV PGG IV OGG II OGG III OAG II

*

** **

*

Fig.2

n=5n=10

OAG III

norm

aliz

ed G

DF

-15 e

xpre

ssio

nFigure(2)

B

D24

7MG

A17

2

LN

-308

LN

-18

U87

MG

D24

7MG

A17

2

LN-2

29

LN-4

28

LN-3

19

LN

-308

U37

3MG

LN

-18

U87

MG

A

norm

aliz

ed G

DF

-15 e

xpre

sion

(arb

itra

ry u

nit

s)

0

200

400

600

800

1000

1200

1400

1600

1800

2000

2200

endogen

ous

pro

life

rati

on r

ate

(arb

itra

ry u

nit

s)

D24

7MG

A17

2

LN-2

29

LN-4

28

LN-3

19

LN

-308

U37

3MG

LN

-18

U87

MG

Fig. 3

C

D24

7MG

LN

-308

U87

MG

0

2

4

6

8

10

pg

GD

F-1

5/1

0.0

00 c

ells

Figure(3)

LN-18 D247MG

AB

LN-308

LN

-31

9 ce

lls are

no

n-in

va

sive

in v

itro

LN-319LN-428U373MG

pre-IS

anti-SGD

F-15

pre-ISanti-SGD

F-15

LNT-229 A172LN-308LN-428U87MGU373MG

pre-IS

anti-SGD

F-15

pre-ISanti-SGD

F-15

LN-229

Fig

. 4

*

Figure 4

Table I. Tumor number, passage number of primary cell cultures, sex and age of patients. Cell cultures corresponding to tumors used for immunohistochemistry are colour-coded.

primary cultures Passage Sex Age

OGG II

NCH171 5 f 53NCH177 5 m 39NCH216 3 m 78NCH240 3 f 64NCH362 6 f 27OGG III

NCH238 3 f 62NCH273 4 m 67NCH281 3 f 59NCH293 4 f 49NCH345 3 m 43OAG IIINCH200 4 m 33NCH248 4 m 27NCH259 4 f 52NCH297 2 f 45NCH349 6 m 38

OAG II

NCH91 13 f 60NCH180 4 f 49NCH204 2 m 40NCH302 2 m 52NCH327 4 m 42

DAG II

NCH110 13 m 25NCH169 7 f 28NCH295 3 m 38NCH323 3 m 25NCH353 9 f 38NCH232 2 m 29NCH170 3 m 63NCH159 2 f 34NCH382 2 m 31NCH185 3 f 35

AAG III

NCH38 11 m 49NCH133 8 m 26NCH158 4 m 65NCH192 3 f 55NCH218 3 f 43NCH389 2 m 65NCH78 5 m 38NCH128 5 m 33NCH326 3 m 31NCH209 6 f 41

SGG IV

NCH173 3 m 18NCH193 4 m 45NCH199 6 m 38NCH214 4 m 40NCH246 2 m 38NCH407 3 f 43NCH203 4 f 58NCH198 2 f 27NCH217 4 m 28PGG IVNCH82 27 m 56NCH89 3 f 41NCH125 3 m 43NCH156 6 m 54NCH149 5 m 64NCH186 3 f 62NCH287 2 m 55NCH210 6 m 60NCH325 7 m 31NCH356 8 f 76tumorsOGG IIINCH273 m 67AAG IIINCH218 f 43NCH326 m 31SGG IVNCH198 f 27PGG IVNCH175 f 38NCH238 f 43NCH255 m 65NCH287 m 55NCH325 m 31NCH334 f 54NCH336 m 66NCH356 f 76NCH357 f 51

Table(1)

Table II. Origin and p53 status of the nine human glioblastoma cell lines used in our study.

cell line origin p53 status

A172 GB wt

D247MG gliosarcoma wt

LN-18 de novo GB mutant / wt

LN-229 de novo GB wt

LN-308 de novo GB null

LN-319 de novo AA mutant

LN-428 de novo GB mutant

U87MG de novo GB wt

U373MG de novo GB mutant

GB: glioblastoma; AA: anaplastic astrocytoma.

Table(2)

Table III. Number of TUNEL-positive cells / 1000 cells in ten glioblastoma cell lines treated

as indicated. Neither GDF-15 anti-serum nor recombinant human (rh)GDF-15 affect apoptosis

in the glioblastoma cell lines under study.

cell line pre-immuneserum

GDF-15 anti-

serum

rhGDF-15

A172 1 1 <1

D247 1 1 <1

LN-18 1 2 2

LN-229 1 2 1

LN-308 2 3 1

LN-319 4 5 4

LN-428 3 3 2

U87MG 3 4 4

U373MG 4 4 6

Table(3)