LAB INVESTIGATION - HUMAN/ANIMAL TISSUE

HMGB1 as an autocrine stimulus in human T98G glioblastomacells: role in cell growth and migration

Rosaria Bassi Æ Paola Giussani Æ Viviana Anelli Æ Thomas Colleoni ÆMarco Pedrazzi Æ Mauro Patrone Æ Paola Viani Æ Bianca Sparatore ÆEdon Melloni Æ Laura Riboni

Received: 19 June 2007 / Accepted: 19 October 2007 / Published online: 2 November 2007

� Springer Science+Business Media, LLC. 2007

Abstract HMGB1 (high mobility group box 1 protein) is

a nuclear protein that can also act as an extracellular trigger

of inflammation, proliferation and migration, mainly

through RAGE (the receptor for advanced glycation end

products); HMGB1–RAGE interactions have been found to

be important in a number of cancers. We investigated

whether HMGB1 is an autocrine factor in human glioma

cells. Western blots showed HMGB1 and RAGE expres-

sion in human malignant glioma cell lines. HMGB1

induced a dose-dependent increase in cell proliferation,

which was found to be RAGE-mediated and involved the

MAPK/ERK pathway. Moreover, in a wounding model, it

induced a significant increase in cell migration, and RAGE-

dependent activation of Rac1 was crucial in giving the

tumour cells a motile phenotype. The fact that blocking

DNA replication with anti-mitotic agents did not reduce the

distance migrated suggests the independence of the pro-

liferative and migratory effects. We also found that glioma

cells contain HMGB1 predominantly in the nucleus, and

cannot secrete it constitutively or upon stimulation;

however, necrotic glioma cells can release HMGB1 after it

has translocated from the nucleus to cytosol. These findings

provide the first evidence supporting the existence of

HMGB1/RAGE signalling pathways in human glioblas-

toma cells, and suggest that HMGB1 may play an

important role in the relationship between necrosis and

malignancy in glioma tumours by acting as an autocrine

factor that is capable of promoting the growth and migra-

tion of tumour cells.

Keywords ERK1/2 � HMGB1 � Human gliomas �Motility � Necrotic death � Proliferation � Rac1 �RAGE

Introduction

The high mobility group box 1 protein (HMGB1) is a

nuclear protein that binds DNA, stabilises nucleosomes,

and is crucial for proper transcriptional regulation [1]. It

has recently been discovered that it is a multifunctional

molecule that can also act as an extracellular trigger and/or

modulator of crucial cell processes such as inflammation,

proliferation, migration and survival [2–4]. HMGB1 can be

present extracellularly after active secretion (particularly

by inflammatory cells) or passive release by necrotic cells

[5–8] and, once in the extracellular milieu, can act as a

paracrine/autocrine factor via its interactions with recep-

tors, mainly RAGE (receptor for advanced glycation end

products) [4, 9]. These interactions trigger the activation of

key signalling pathways involved in the regulation of cell

differentiation, growth, motility and death [9, 10].

One important aspect of the multifunctional role of

HMGB1 is the fact that it and RAGE are expressed in a

variety of cancers and tumour cell lines, and it has been

Rosaria Bassi and Paola Giussani contributed equally to this study.

R. Bassi � P. Giussani � V. Anelli � T. Colleoni � P. Viani �L. Riboni (&)

Department of Medical Chemistry, Biochemistry and

Biotechnology, University of Milan, L.I.T.A.-Segrate,

Via Fratelli Cervi 93, 20090 Segrate, Milan, Italy

e-mail: laura.riboni@unimi.it

M. Pedrazzi � B. Sparatore � E. Melloni

Department of Experimental Medicine and Center of Excellence

for Biomedical Research, University of Genoa, Genoa, Italy

M. Patrone

Department of Environmental and Cell Life, University

of Eastern Piedmont, Alessandria, Italy

123

J Neurooncol (2008) 87:23–33

DOI 10.1007/s11060-007-9488-y

suggested that they are both involved in regulating tumour

properties [4, 10, 11]. In most human tumours, HMGB1–

RAGE interactions have been extensively related to tumour

growth and metastasis, and poor patient survival [10–12],

although the precise mechanism of action remains

unknown. Furthermore, some studies have found that de-

regulated RAGE expression is associated with more

aggressive lung tumours [13, 14], and contributes to neo-

plastic transformation in myoblasts [15]. In general, it

seems that the role of HMGB1–RAGE in cancer may

depend on the type of tumour cell.

Notwithstanding this evidence, and the fact that

HMGB1 has been proposed as a novel target for cancer

therapy [16], no data have yet been published concerning

its expression and functional role (and those of RAGE) in

human astrocytomas, the most frequent and deadly primary

intracranial tumours in adults. These tumours can show

different stages of malignancy, and most low-grade

tumours subsequently progress [17]. The characteristics of

high-grade tumours (including anaplastic astrocytomas and

glioblastomas) are excessive proliferation, infiltrative

growth, increased angiogenesis and resistance to apoptosis,

which limit the success of current therapeutic approaches.

Malignant gliomas also show areas of necrosis, which is a

significant prognostic factor and allows the categorisation

of a tumour as glioblastoma [17]. It has been suggested that

the altered regulation of the triggers and modulators of cell

growth, motility and death plays a significant role in gli-

oma progression and invasion [18, 19].

Given that the interaction between HMGB1 and RAGE

may promote the proliferation and invasion of tumour cells

[4, 10, 11], we sought to determine whether either or both

may play a direct role in glioma cell growth and motility.

Our findings provide the first evidence of the existence of

HMGB1/RAGE signalling pathways in human glioblas-

toma cells, and suggest that the HMGB1 released from

tumour cells undergoing necrosis may play a critical role in

glioma progression.

Material and methods

Material

Dulbecco’s modified Eagle medium (DMEM), foetal calf

serum (FCS), bovine serum albumin (BSA), phorbol-12-

myristate-13-acetate (PMA), dibutyryl cAMP (dbcAMP),

lipopolysaccharide (LPS), Hoechst 33342 and C2-ceramide

(N-acetyl-D-erythro-sphingosine) (C2-Cer) all came from

Sigma-Aldrich (Milan, Italy). Eukaryotic recombinant

HMGB1 was obtained and purified to homogeneity as

previously described [20]. The amount of LPS in the

HMGB1 preparations was 2.5–3 pg/lg protein. The anti-

HMGB1 monoclonal antibody was prepared as previously

described [21].

Cell cultures and treatments

We used two human glioma cell lines derived from an

anaplastic astrocytoma (CCF-STTG1) and a glioblastoma

(T98G), both of which were purchased from the American

Tissue Culture Collection (Rockville, MD), and cultured as

recommended by the supplier. The cells were grown at

37�C in DMEM supplemented with 10% FCS, 1 mM Na-

pyruvate, 2 mM L-glutamine, 100 U/ml penicillin, 100 lg/

ml streptomycin and 0.25 lg/ml amphotericin B in a fully-

humidified incubator containing 5% CO2 and 95% air. The

cells were starved of serum before the treatments, and

thereafter kept under serum-free conditions. Eukaryotic

recombinant HMGB1 was added to the cells in DMEM

containing 10 lg/ml fatty acid-free BSA (DMEM +

BSA), and the cells were incubated for different times. The

human monoclonal anti-RAGE antibody (10 lg/ml) raised

against the extracellular domain of RAGE (AF1145, R&D

Systems, MN) was used to block the binding of HMGB1 to

RAGE. C2-Cer (1–50 lM) was administered in serum-free

DMEM as previously described [22].

Immunoblot analyses

HMGB1 expression was determined in pre-confluent cells,

after lysis with 5% glycerol in 20 mM Tris–HCl, pH 8.5,

1 mM EDTA, 1 mM DTT, 10 lg/ml aprotinin, 5 lg/ml

leupeptin, 5 lg/ml pepstatin and 3 lg/ml bestatin. The cell

lysates (1–5 lg protein) and recombinant HMGB1 (10–

50 ng) were separated by means of 12.5% SDS-PAGE

followed by immunoblotting with 0.18 lg/ml monoclonal

anti-HMGB1 antibody and 1:10,000 HRP-conjugated anti-

mouse antibody (Santa Cruz Biotechnology, Inc. CA).

RAGE expression was evaluated after cell lysis with 1%

Triton X-100 in 10 mM Tris–HCl, pH 7.5, 150 mM NaCl,

2 mM EDTA, 2 mM Na-orthovanadate, 1 mM phenyl-

methylsulfonyl fluoride (PMSF), 2 lg/ml aprotinin and

2 lg/ml leupeptin. After sonication, the cell lysates were

centrifuged 10,000 9 g for 10 min, and the obtained

supernatants (25–50 lg as cell proteins) underwent 10%

SDS-PAGE followed by immunoblotting using 1:1,000

mouse anti-RAGE (MAB5328, Chemicon International,

Temecula, CA) and 1:5,000 HRP-conjugated anti-mouse

antibody (Santa Cruz Biotechnology, Inc. CA).

In order to detect extracellular signal-regulated kinase

(ERK) 1/2 phosphorylation, the cell extracts (10–30 lg as

protein) were analysed by means of Western blotting

essentially as previously described [23]. In brief, the

24 J Neurooncol (2008) 87:23–33

123

glioma cells were lysed with 1% Triton X-100 in 20 mM

Tris–HCl, pH 7.4, 137 mM NaCl, 2 mM EDTA, 2 mM

sodium pyrophosphate, 50 mM b-glycerophosphate, 1 mM

sodium orthovanadate, 1 mM PMSF, and 2 lg/ml each of

aprotinin and leupeptin for 10 min at 4�C. After sonication,

the cell lysates were centrifuged at 13,000g for 10 min, and

the proteins (corresponding to 10–20 lg) were separated

on 10% SDS-PAGE, and blotted with 1:500 phospho-

specific antibodies against phosphorylated ERK1/2 and

phosphorylation-independent anti-ERK2 antibodies (Santa

Cruz Biotechnology, Inc. CA). Mouse secondary antibod-

ies (1:5,000) were HRP-conjugated (Santa Cruz

Biotechnology, Inc. CA).

The immune complexes were visualised using an

enhanced chemiluminescence detection kit (Pierce Bio-

technology, Inc., IL), and their optical densities were

assessed by scanning the autoradiograms (BIORAD GS-

700 imaging densitometer equipped with Quantity One

software). The proteins were assayed according to Lowry

et al. [24], using BSA as standard.

Evaluation of extracellular HMGB1

Extracellular HMGB1 was determined essentially as

recently described [21]. Briefly, the cells were treated with

10 lg/ml heparin (5 min), and the conditioned media were

collected and clarified by centrifugation (13,000 9 g, for

10 min). Triton X-100 (0.1% final concentration) was then

added. The cells were lysed with 1% Triton X-100 in

DMEM + BSA, followed by 1:10 dilution in the same

medium. Aliquots of the media and cells underwent protein

precipitation in 10% trichloroacetic acid (TCA), followed

by 12.5% SDS-PAGE separation, and immunoblotting with

0.18 lg/ml anti-HMGB1 antibody and 1:10,000 HRP-

conjugated anti-mouse antibody (Santa Cruz Biotechnol-

ogy, Inc. CA). The optical density of the HMGB1 bands

was analysed by densitometry (BIORAD GS-700, Quantity

One software). Different amounts (5–100 ng) of recombi-

nant HMGB1 were added to DMEM + BSA and

processed in parallel, and the related densitometric results

were used to calculate the concentration of released

HMGB1.

Immunolocalisation of HMGB1

In order to image HMGB1, the cell cultures were washed

thoroughly three times with phosphate buffered saline

(PBS; 10 mM sodium phosphate, 137 mM NaCl and

2.7 mM KCl, pH 7.4), and the cultures were fixed for

10 min with solutions of 2% paraformaldehyde in PBS and

permeabilised by 0.5% Triton X-100 in PBS. Non-specific

interactions were blocked by means of 30 min incubation

in 0.5% Triton X-100, 5% BSA, and 1% FCS in PBS. The

cells were incubated with 1:200 anti-HMGB1 antibody

[21] in PBS containing 0.5% Triton X-100, 5% BSA and

1% FCS, and then with the corresponding FITC-conjugated

secondary antibody (1:200) (Invitrogen-Molecular Probes).

Staining was acquired using a fluorescence microscope

(Olympus BX-50) equipped with a high-resolution charge-

coupled device camera and image analysis software (Col-

orview 12 and Soft Imaging System GmbH).

Pull-down assay for Rac1

The levels of GTP-bound Rac1 were measured in a pull-

down assay using an EZ-DetectTM Rac1 activation kit

(Pierce Biotechnology, Inc., IL) in accordance with the

manufacturer’s protocol. The same amounts of cell lysates

loaded with GTPcS or GDP were included in each exper-

iment and respectively used as positive and negative

controls. Equal loading was confirmed by detecting total

Rac1 in the cell lysates.

Proliferation assays

Cell proliferation was measured by means of the [3H]thy-

midine uptake assay as previously described [22]. Briefly,

the cells were incubated in DMEM containing 0.5% FCS

for 72 h, and the quiescent cells were then treated with 2–

50 nM HMGB1 in serum-free DMEM + BSA for 24 h,

and pulsed with 1 lCi/ml of [3H]thymidine (PerkinElmer,

Inc., Waltham, MA) for 4 h before harvesting. Finally, the

amount of radioactivity incorporated into DNA was

determined [22].

Evaluation of cell migration

A wound healing assay was used to evaluate cell migration.

Confluent T98G cells were serum-starved for 24 h, and

then a scratch was made with a micropipette tip to create a

cell-free area. The scratched monolayers were treated with

HMGB1, and the images were captured 0 and 24 h after

wounding using an Olympus IX-50 microscope equipped

with a Variocam camera. Cell migration distance was

determined by measuring the wound width and subtracting

this from the initial value. A total of twelve areas were

randomly selected in each well, and the cells in three wells

of either group were quantified in each experiment.

To determine the contribution of cell proliferation to the

measured migration distance, the cells were incubated with

J Neurooncol (2008) 87:23–33 25

123

the anti-mitotic drug 5-fluorouracil (5-FU) (50 lg/ml) or

aphidicolin (AP) (20 lg/ml) for 24 h.

Cell viability assays

Cell viability was assessed by means of the trypan-blue

exclusion test and lactate dehydrogenase (LDH) activity

assays. LDH was measured in conditioned media and cells

using a LDH assay (TOX8 Kit, Sigma-Aldrich, Italy). The

morphological features of apoptosis (DNA condensation

and/or apoptotic bodies) were visualised using the blue

nuclear fluorescence induced by the Hoechst 33342

supravital DNA stain (5 lg/ml for 15 min at 37�C). The

cells were observed through an Olympus BX-50 fluores-

cence microscope.

Data analysis

The data are expressed as mean values ± SD, and are

representative results from at least three independent

experiments performed using a separate culture prepara-

tion. Statistical significance was determined by means of

the two-tailed unpaired Student’s t-test.

Results

Effect of HMGB1 on cell proliferation and ERK1/2

phosphorylation

A number of studies have revealed that extracellularly

added HMGB1 stimulates several cell lines and induces

responses related to inflammation, proliferation and

migration [2–4]. We evaluated the functioning and

responsiveness of glioma cells by first investigating the

capacity of exogenous HMGB1 to promote proliferation.

Quiescent T98G cells were treated with increasing con-

centrations of HMGB1, and [3H]thymidine incorporation

into DNA was assayed as the index of cell duplication. As

shown in Fig. 1a, HMGB1 2–50 nM added to quiescent

T98G cells significantly stimulated DNA synthesis in a

dose-dependent manner.

To study the signalling pathway mediating HMGB1-

induced proliferation, we investigated the effect of

HMGB1 on ERK1/2, the MAPK signalling modules that

play a major role in cell proliferation [25] and tumori-

genesis [26]. In quiescent cells, HMGB1 addition led to

dose-dependent and rapid increase in ERK phosphoryla-

tion, which returned to the original value after 60–120 min

of incubation (Fig. 1b and c). We also found that HMGB1-

induced ERK activation and cell proliferation were both

effectively inhibited by PD98059, the selective inhibitor of

the mitogen-activated protein kinase (MEK) (Fig. 2a and

b). In addition, functional blockade of RAGE by means of

neutralising anti-RAGE antibody strikingly prevented both

the ERK phosphorylation and cell proliferation elicited by

exogenous HMGB1 (Fig. 2a and b).

Effect of HMGB1 on cell migration and Rac1

activation

A number of studies suggest that HMGB1 plays a key role

as a regulator of invasive migration into tumour cells [10,

27] and this, together with the highly infiltrative properties

of human gliomas, led us to investigate the ability of

HMGB1 to induce glioma cell motility. Using a scratch

wound healing assay, we found that exposure of T98G cells

to HMGB1 stimulated glioma cells to extend membrane

protrusions into the wounded areas, and led to a significant

and dose-dependent increase in the rate of cell invasion of

the artificial wound (Fig. 3a and b). A similar pro-migra-

tory effect was also seen when HMGB1 was added to CCF-

STTG cells (data not shown). In both cell types, the anti-

mitotic agents AP and 5-FU effectively inhibited HMGB1-

induced cell growth (not shown), but not HMGB1-induced

cell migration (Fig. 3c). We also found that HMGB1-

induced cell migration was abated when cell stimulation

was elicited in the presence of RAGE-blocking antibody,

but not in the presence of irrelevant control IgG (Fig. 3d).

It is well known that the activated form of Rac1, a

member of the Rho-family of small GTPases, stimulates

cell migration and, as recent studies have revealed that it

plays a major role in glioma invasion [28], we investigated

whether HMGB1 promotes activation of Rac1 in T98G

cells. HMGB1 addition increased the levels of GTP-bound

Rac1, whose activation was substantial within 2 min of

HMGB1 exposure, peaked at 10 min, and declined at

30 min (Fig. 4a). This event was also suppressed in the

presence of neutralising anti-RAGE antibody, but not in the

presence of irrelevant IgG (Fig. 4b). These data show that

RAGE binding induces Rac1 activation and concomitantly

promotes cell migration, and thus support the view of

RAGE as the major receptor mediating HMGB1-dependent

migration [9, 10].

HMGB1 and RAGE expression in human glioma cell

lines

The expression of HMGB1 protein and its receptor RAGE

has been detected in many forms of cancer and different

tumour cell lines [4, 11, 12], but nothing is known about

their expression in human glioma cells. Western blots of

26 J Neurooncol (2008) 87:23–33

123

cell lysates using a monoclonal anti-HMGB1 antibody

revealed similar amounts of HMGB1, corresponding to

0.8–1.0% of total cell protein, in both the CCF-STTG1 and

Fig. 1 Effects of HMGB1 on cell proliferation and ERK1/2 activation.

Quiescent T98G cells were treated with (a) 0–50 nM HMGB1 in serum-

free DMEM for 24 h; (b) 0–50 nM HMGB1for 30 min; or (c) 10 nM

HMGB1for0–120 min,andthenretrieved.(a)3H-thymidineincorporation

intoDNA,and(bandc)ERK1/2activationweredeterminedasdescribedin

‘‘Materialandmethods’’; (b)and(c) showarepresentative immunoblotand

themeanlevelsofphosphorylatedERK1/2 (p-ERK1/2).Meanpercentages

(±SD) of HMGB1-unstimulated cells (100%) of at least three independent

experiments performed in duplicate. **: P \0.01 vs. unstimulated cells

Fig. 2 Effects of MEK inhibition and anti-RAGE neutralising

antibody on HMGB1-induced ERK1/2 activation and cell prolifera-

tion. The cells were pretreated for 2 h with 50 lM PD98059, 10 lg/

ml anti-RAGE neutralising antibody, or 10 lg/ml irrelevant IgG, and

then stimulated with 50 nM HMGB1 as shown in (a). Phosphorylated

ERK levels (b) and DNA synthesis (c) were evaluated after 30 min

and 24 h, respectively. Mean values ± SD of three independent

experiments. **: P \ 0.01 vs. corresponding cells not treated with

HMGB1 (taken as 100%)

J Neurooncol (2008) 87:23–33 27

123

T98G cell lines (Fig. 5a). RAGE protein was also expres-

sed in both, but to a greater extent in T98G (Fig. 5b).

Extracellular release of HMGB1

In order to examine the potential secretion/release of

HMGB1 by human glioma cells, we sought HMGB1 pro-

tein in the medium after various conditions of glioma cell

incubation. The experiments were performed using both

T98G and CCF-STTG1 cells but, as the results were

identical, we shall limit our description to the T98G cells.

After the cells had been incubated in serum-free medium

for 2–16 h, immunoblotting detected HMGB1 in the

extracellular milieu of T98G cells in a time-dependent

manner (Fig. 6a). The amount of HMGB1 protein in the

extracellular environment was less than 7% of cellular

HMGB1 even after 16 h of incubation, and extracellular

LDH levels (evaluated as an index of cell damage) were

similar to those of HMGB1 in both control and treated cells

(Fig. 6b). Further experiments showed that the amount of

HMGB1 in the extracellular milieu was not affected by cell

treatments with known inducers of HMGB1 secretion, such

as dBcAMP, PMA or LPS (Fig. 6c) and, once again, the

Fig. 3 Effects of HMGB1 on cell migration. Confluent T98G cell

monolayers were serum-starved overnight and underwent in vitro

wound-healing assays in the absence or presence of different doses of

HMGB1. (a) Phase-contrast images of wound edges 0 (upper panels)

and 24 h (lower panels) after monolayer wounding in the absence

(control) or presence (HMGB1) of 50 nM HMGB1. Bar: 100 lm; (b)

Quantitative results for the migration of T98G cells in the presence of

HMGB1. The cells were allowed 24 h to migrate in the absence

(control) or presence of increasing concentrations of recombinant

HMGB1 in serum-free medium; cell migration distance was deter-

mined as described in ‘‘Material and methods’’. Mean percentages

(±SD) of control without HMGB1 from at least three independent

experiments. **: P \ 0.01 vs. cells not treated with HMGB1. (c)

Quantitative results for the migration of T98G cells incubated for

24 h with (black bars) or without (white bars) 50 nM HMGB1 in the

absence or presence of aphidicolin (AP, 20 lg/ml) or 5-fluorouracil

(5-FU, 50 lg/ml). Averages (±SD) of three independent experiments.

**: P \ 0.01 vs. cells not treated with HMGB1. (d) Quantitative

migration of T98G cells incubated for 24 h with (black bars) or

without (white bars) 50 nM HMGB1 in the absence or presence of

10 lg/ml neutralizing anti-RAGE antibodies or irrelevant IgG. Mean

values ± SD of at least three independent experiments. **: P \ 0.01

vs. cells not treated with HMGB1

28 J Neurooncol (2008) 87:23–33

123

levels of released LDH were similar to those of HMGB1

(data not shown). The results of this set of experiments

suggested that human glioma cells cannot release HMGB1

spontaneously or upon stimulation.

Recent studies indicate that necrotic cells can release

HMGB1 into the extracellular milieu [29, 30] and so, as

necrosis is a characteristic feature of malignant gliomas

[17], we evaluated whether glioma cells undergoing

necrotic death might be a source of extracellular HMGB1.

To do so, we treated them with ceramide, an important cell

messenger involved in triggering apoptotic/necrotic pro-

cesses in many cell types including cancer cells [31, 32],

the use of which was also prompted by previous studies

showing that treating T98G cells with the cell-permeable

ceramide analogue C2-Cer induces non-apoptotic, necrotic

cell death [33]. Sixteen hours’ treatment with 10–50 lM

C2-Cer led to dose-dependent cell toxicity (Fig. 7a), and a

dramatic increase in LDH activity in the medium (Fig. 7b).

As in previous studies [33], we did not observe the typical

morphological features of apoptosis, such as cell shrinkage

and separation from neighbouring cells. Under these con-

ditions, a substantial amount of HMGB1 was found in the

extracellular milieu, which increased with C2-Cer doses

and paralleled extracellular LDH levels (Fig. 7c). We

estimated that the levels of HMGB1 in the medium after

16 h exposure to 10 and 50 lM C2-Cer were respectively

18.2 ± 2.6 and 82.7 ± 8.8 nM.

Fig. 4 Effects of HMGB1 and HMGB1–RAGE interactions on Rac1

activation and cell migration. The cells were incubated with

recombinant HMGB1 for the indicated times, and the amounts of

GTP-bound and total Rac1 were detected by immunoblotting.

Relative Rac1 activation was determined on the basis of the amount

of GTP-bound Rac1 normalised to the amount of Rac1 in the cell

lysates, and the value from control. The value of the cells not treated

with HMGB1 was arbitrarily set at 1. Mean values ± SD of three

independent experiments. (a) Representative immunoblot of GTP-

Rac1 and total Rac1, and relative amount of GTP-Rac1 after 0–

30 min treatment with 50 nM HMGB1. (b) Representative immuno-

blot of GTP-Rac1 and relative amount of GTP-Rac1 after and 10 min

incubation without (1) or with (2–4) 50 nM HMGB1 in the absence

(1, 2) or presence of neutralising anti-RAGE antibodies (10 lg/ml)

(3) or irrelevant control IgG (4). **: P \ 0.01 vs. cells not treated

with HMGB1

Fig. 5 HMGB1 and RAGE expression in human glioma cells.

Western blotting of cell lysates from two different dishes of CCF-

STTG1 (1) and T98G cells (2). Panel (a) HMGB1 immunodetection,

2 lg as cell proteins, 12.5% SDS-PAGE. St, standard HMGB1, 50 ng;

panel (b) RAGE immunodetection, 25 lg as cell proteins, 10% SDS-

PAGE. The blots were reprobed with an anti-GAPDH antibody to

ensure equivalent protein loading. A representative blot of three

J Neurooncol (2008) 87:23–33 29

123

Although HMGB1 is mainly a nuclear, chromosome-

associated protein, its passive release with necrosis is

believed to occur as a result of passive diffusion from its

cytoplasmic pool [29]. We therefore investigated whether

C2-Cer treatment affects its subcellular localization. After

immunofluorescent staining with an anti-HMGB1 mono-

clonal antibody [20], a large number of control cells

showed remarkable HMGB1 immunoreactivity in the

nucleus, but this was more diffusely distributed in both the

nuclei and cytosol of C2-Cer treated cells (Fig. 7d). In

addition, and consistent with the findings of a previous

study [33], Hoechst 33342 staining of the C2-Cer-treated

cells revealed no nuclear condensation or blebbing

(Fig. 7d). Notably, the change in HMGB1 distribution was

already clear after only 8 h treatment (Fig. 7d): i.e. prior to

cell death as determined by trypan blue exclusion (not

shown). Taken together, these findings indicate that C2-Cer

induces necrotic cell death, and that dying cells can relo-

cate and release HMGB1 upon death.

Discussion

It has been suggested that the interaction of HMGB1 and

RAGE promotes the proliferation and invasion of various

tumour cells [4, 10]. The main new finding of our study is

that HMGB1 can operate as an autocrine factor that is

capable of stimulating the growth and migration of human

glioma cells. To the best of our knowledge, this is the first

experimental evidence showing that HMGB1 plays a bio-

logical role in human glioma cells.

First of all, extracellular HMGB1 plays a key role in

promoting the proliferative properties of human T98G

glioblastoma cells as the administration of nanomolar

amounts to quiescent cells significantly stimulated cell

growth. The mitogenic effect of HMGB1 seemed to be

mainly mediated by RAGE (also expressed by human

glioma cells) insofar as its functional blockade by means of

specific RAGE neutralising antibody strikingly prevented

the cell proliferation elicited by exogenous HMGB1.

When studying the signalling pathways involved in this

effect, we found that a key role is played by the ERK1/2

MAP kinase pathway, a central regulator of cell prolifer-

ation [34] that is commonly activated in gliomas [35]. The

HMGB1 treatment of glioma cells led to ERK1/2 phos-

phorylation, which is crucial to HMGB1-induced cell

growth. Recent in vitro binding studies using human

RAGE mutants have identified a cytoplasmic region of

RAGE as an ERK docking site [36], thus suggesting that

ERK signalling occurs as a result of direct ERK-RAGE

interactions. Although this interaction cannot be excluded

in glioma cells, the fact that the MEK inhibitor PD98059

almost completely prevented the ERK phosphorylation and

cell proliferation induced by HMGB1 supports the

hypothesis that HMGB1 regulates the proliferation of

T98G glioma cells through RAGE/MEK/ERK signalling.

Fig. 6 Extracellular release of HMGB1 by human glioma cells.

Subconfluent T98G cells were incubated in serum-free DMEM for

different times, and HMGB1 in the medium and cells was evaluated

by immunoblotting. The figures show a representative immunoblot of

HMGB1 from the medium and cell lysate (a), and extracellular levels

of HMGB1 and LDH (b). Mean percentages (±SD) of total (cellular

plus extracellular) content. (c) Subconfluent T98G cells were

incubated in serum-free DMEM in the absence (control) or presence

of dBcAMP (0.5 mM), PMA (0.1 mM), or LPS (0.1 lg/ml). The cell

lysates and medium were collected after 16 h. The figures show a

representative immunoblot of HMGB1 from the medium and cell

lysate. The numbers at the top of the lanes refer to the percentage

extracellular HMGB1, evaluated as above. Similar results were

obtained in two other experiments

30 J Neurooncol (2008) 87:23–33

123

Another important finding is that the HMGB1-induced

effects on glioma cells are not restricted to mitosis but

extend to motility through enhanced migration. A mono-

layer scratch wound healing assay showed that HMGB1

markedly stimulated cell motility, and this was unaffected

by antimitotic agents. The pro-migratory effect induced by

HMGB1 therefore seems to be distinct from its prolifera-

tive effect, and consistent with the observation that the

proliferation and migration of astrocytoma cells are

dichotomous behaviours [37].

The migratory effect of HMGB1 on glioma cells was

suppressed in the presence of the neutralising anti-RAGE

antibody, thus indicating that the action of HMGB1 on its

RAGE membrane receptor plays a key role in giving

tumour cells a motile phenotype, in agreement with pre-

vious studies of other cell types [10, 11].

It has been reported that the Rac1 member of the Rho

family of small GTPases is a major intracellular signalling

pathway of RAGE, whose phosphorylation leads to inte-

gration of the cytoskeleton and the formation of

cytomembranous ruffling, resulting in activated cell

mobility [38]. We found that Rac1 plays a critical role in

the HMGB1-stimulated migration of glioma cells and that

HMGB1 can induce cell motility via the RAGE/Rac1

pathway. It is worth noting that Rac1 is essential for var-

ious aspects of malignant transformation, including tumour

cell invasion [39, 40] and, in glioma cells, the depletion of

Rac1 expression by siRNA oligonucleotides leads to a

decrease in cell migration and invasion [41, 42].

Further information came from our experiments in

which we investigated glioma cells as a possible source of

extracellular HMGB1. We first found that, although human

Fig. 7 Extracellular release of

HMGB1 by necrotic cells.

Subconfluent T98G cells were

treated with 0–50 lM C2-Cer in

serum-free DMEM. After 16 h

incubation, cell death (a), and

the extracellular levels of LDH

(b) and HMGB1 (c), were

evaluated as described in

‘‘Material and methods’’.

Average values ± SD of at least

three independent experiments.

**: P \ 0.01 vs. cells not

treated with HMGB1. The upper

part of c also shows a

representative Western blot of

HMGB1 in the culture

supernatant. (d)

Immunocytochemistry

visualising the HMGB1 pattern

(upper panels) after 8 h

treatment of T98G cells with

50 lM C2-Cer (C2-Cer) or no

treatment (control). The cell

nuclei were stained blue with

Hoechst 33342 (lower panels).

Similar patterns were obtained

after 16 h treatment and in two

other experiments

J Neurooncol (2008) 87:23–33 31

123

T98G and CCF-STTG1 glioma cells constitutively express

HMGB1, they do not seem to be able to secrete it consti-

tutively or upon stimulation. This observation conflicts

with the finding that a number of other tumour cells, such

as erythroleukemia [43], neuroblastoma [38] and colon

cancer cells [44], can release HMGB1 extracellularly, and

supports the view that not all cancer cells actively secrete

HMGB1 [45]. Immunocytochemical analysis of T98G cells

showed that the cellular pattern of HMGB1 was predomi-

nantly nuclear and, although the functional role of nuclear

HMGB1 in glioma cells is unknown, it is tempting to

speculate that the anti-apoptotic properties of HMGB1 [46]

may contribute to the apoptosis resistance of malignant

gliomas.

Further experiments showed that human glioma cells

undergoing ceramide-induced necrotic death release large

amounts of HMGB1 into extracellular space, thus sup-

porting the hypothesis that HMGB1 is an extracellular

signal of necrosis [8]. This release was probably passive

insofar as it was associated with cell death and strictly

correlated with LDH release.

Although HMGB1 is mainly a nuclear, DNA-associated

protein, it is believed that its release by necrotic cells

occurs passively from the cytoplasmic pool of the protein

[29]. We found that its release by necrotic glioma cells was

preceded by its translocation from the nucleus to the

cytoplasm, which suggests that it may be related to

HMGB1 relocation [30]. The mechanism by which

HMGB1 is released from chromatin during necrosis is

largely unknown, although one very recent study [47] has

found that the activation of poly(ADP-ribose) polymerase

(PARP) is crucial to this release and therefore to the

translocation of HMGB1 from the nucleus to the cytosol.

Interestingly, it has been demonstrated that poly(ADP-

ribosyl)ation occurs in association with necrosis in human

glioblastomas as well as in multicellular tumour spheroids

derived from human glioma cell lines [48].

The development of necrosis is a characteristic feature

of human glioblastomas [17], and clinical studies have

correlated the degree of necrosis with a worse prognosis

[49], thus implying its importance in the pathogenesis of

glioblastomas. Our finding that human glioma cells can

respond to HMGB1 and release HMGB1 at the time of

necrotic death is intriguing as it suggests that glioma-

associated necrosis in vivo may have tumour-promoting

features. In this context, necrotic glioma cells releasing

HMGB1 would benefit neighbouring glioma cells by pro-

moting their proliferative and migratory behaviour.

In brief, our data suggest that HMGB1 may contribute to

glioma progression by acting as an autocrine factor capable

of promoting the growth and migration of tumour cells, and

may play an important role in modulating the relationships

between necrosis and malignancy in glioma tumours.

Further studies using additional models of gliomas

which closely mimic the human tumour, such as short term

cell cultures growing as neurospheres, as well as investi-

gating HMGB1/RAGE in glioma samples in relation to

necrotic areas are needed to support and strengthen the

biological relevance of our findings. Resolving this issue

could lead to the creation of novel therapeutic approaches

towards the treatment of glioma patients.

Acknowledgements This work was supported in part by grants

from the Italian Ministry of University and Scientific and Techno-

logical Research PRIN and FIRST to L.R.

References

1. Lotze MT, Tracey KJ (2005) High-mobility group box 1 protein

(HMGB1): nuclear weapon in the immune arsenal. Nat Rev

Immunol 5:331–342

2. Muller S, Ronfani L, Bianchi ME (2004) Regulated expression

and subcellular localization of HMGB1, a chromatin protein with

a cytokine function. J Intern Med 255:332–343

3. Yang H, Wang H, Czura CJ, Tracey KJ (2005) The cytokine

activity of HMGB1. J Leukoc Biol 78:1–8

4. Ulloa L, Messmer D (2006) High-mobility group box 1

(HMGB1) protein: friend and foe. Cytokine Growth Factor Rev

17:189–201

5. Bianchi ME (2004) Significant (re)location: how to use chromatin

and/or abundant proteins as messages of life and death. Trends

Cell Biol 14:287–293

6. Erlandsson Harris H, Andersson U (2004) The nuclear protein

HMGB1 as a proinflammatory mediator. Eur J Immunol

34:1503–1512

7. Wang H, Yang H, Tracey KJ (2004) Extracellular role of

HMGB1 in inflammation and sepsis. J Intern Med 255:320–331

8. Raucci A, Palumbo R, Bianchi ME (2007) HMGB1: a signal of

necrosis. Autoimmunity 40:285–289

9. Bierhaus A, Humpert PM, Morcos M, Wendt T, Chavakis T,

Arnold B, Stern DM, Nawroth PP (2005) Understanding RAGE,

the receptor for advanced glycation end products. J Mol Med

83:876–886

10. Huttunen HJ, Rauvala H (2004) Amphoterin as an extracellular

regulator of cell motility: from discovery to disease. J Intern Med

255:351–366

11. Ellerman JE, Brown CK, de Vera M, Zeh HJ, Billiar T, Rubartelli

A, Lotze MT (2007) Masquerader: high mobility group box-1 and

cancer. Clin Cancer Res 13:2836–2848

12. Kuniyasu H, Chihara Y, Takahashi T (2003) Co-expression of

receptor for advanced glycation end products and the ligand

amphoterin associates closely with metastasis of colorectal can-

cer. Oncol Rep 10:445–448

13. Huttunen HJ, Fages C, Kuja-Panula J, Ridley AJ, Rauvala H

(2002) Receptor for advanced glycation end products-binding

COOH-terminal motif of amphoterin inhibits invasive migration

and metastasis. Cancer Res 62:4805–4811

14. Bartling B, Hofmann HS, Weigle B, Silber RE, Simm A (2005)

Down-regulation of the receptor for advanced glycation end-

products (RAGE) supports non-small cell lung carcinoma. Car-

cinogenesis 26:293–301

15. Riuzzi F, Sorci G, Donato R (2006) The amphoterin (HMGB1)/

receptor for advanced glycation end products (RAGE) pair

modulates myoblast proliferation, apoptosis, adhesiveness,

migration, and invasiveness. Functional inactivation of RAGE in

32 J Neurooncol (2008) 87:23–33

123

L6 myoblasts results in tumor formation in vivo. J Biol Chem

281:8242–8253

16. Lotze MT, DeMarco RA (2003) Dealing with death: HMGB1 as

a novel target for cancer therapy. Curr Opin Investig Drugs

4:1405–1409

17. Kleihues P, Louis DN, Scheithauer BW, Rorke LB, Reifenberger

G, Burger PC, Cavenee WK (2002) The WHO classification of

tumors of the nervous system. J Neuropathol Exp Neurol 61:

215–225

18. Tysnes BB, Mahesparan R (2001) Biological mechanisms of

glioma invasion and potential therapeutic targets. J Neurooncol

53:129–147

19. Hulleman E, Helin K (2005) Molecular mechanisms in glioma-

genesis. Adv Cancer Res 94:1–27

20. Sparatore B, Passalacqua M, Patrone M, Melloni E, Pontremoli S

(1996) Extracellular high-mobility group 1 protein is essential for

erythroleukaemia cell differentiation. Biochem J 320:253–256

21. Sparatore B, Patrone M, Passalacqua M, Pedrazzi M, Ledda S,

Pontremoli S, Melloni E (2005) Activation of A431 human car-

cinoma cell motility by extracellular high-mobility group box 1

protein and epidermal growth factor stimuli. Biochem J 389:

215–221

22. Riboni L, Viani P, Bassi R, Giussani P, Tettamanti G (2001)

Basic fibroblast growth factor-induced proliferation of primary

astrocytes. J Biol Chem 276:12797–12804

23. Bassi R, Anelli V, Giussani P, Tettamanti G, Viani P, Riboni L

(2006) Sphingosine-1-phosphate is released by cerebellar astro-

cytes in response to bFGF and induces astrocyte proliferation

through Gi-protein-coupled receptors. Glia 53:621–630

24. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ (1951) Protein

measurement with the Folin phenol reagent. J Biol Chem

193:265–275

25. Rubinfeld H, Seger R (2005) The ERK cascade: a prototype of

MAPK signaling. Mol Biotechnol 31:151–174

26. Wallace EM, Lyssikatos JP, Yeh T, Winkler JD, Koch K (2005)

Progress towards therapeutic small molecule MEK inhibitors for

use in cancer therapy. Curr Top Med Chem 5:215–229

27. Taguchi A, Blood DC, Del Toro G et al (2000) Blockade of

RAGE-amphoterin signalling suppresses tumour growth and

metastases. Nature 405:354–360

28. Salhia B, Tran NL, Symons M, Winkles JA, Rutka JT, Berens

ME (2006) Molecular pathways triggering glioma cell invasion.

Expert Rev Mol Diagn 6:613–626

29. Scaffidi P, Misteli T, Bianchi ME (2002) Release of chromatin

protein HMGB1 by necrotic cells triggers inflammation. Nature

418:191–195

30. Bianchi ME, Manfredi A (2004) Chromatin and cell death.

Biochim Biophys Acta 1677:181–186

31. Mimeault M (2002) New advances on structural and biological

functions of ceramide in apoptotic/necrotic cell death and cancer.

FEBS Lett 530:9–16

32. Ogretmen B, Hannun YA (2004) Biologically active sphingoli-

pids in cancer pathogenesis and treatment. Nat Rev Cancer

4:604–616

33. Mochizuki T, Asai A, Saito N, Tanaka S, Katagiri H, Asano T,

Nakane M, Tamura A, Kuchino Y, Kitanaka C, Kirino T (2002)

Akt protein kinase inhibits non-apoptotic programmed cell death

induced by ceramide. J Biol Chem 277:2790–2797

34. Meloche S, Pouyssegur J (2007) The ERK1/2 mitogen-activated

protein kinase pathway as a master regulator of the G1- to S-

phase transition. Oncogene 26:3227–3239

35. Sanson M, Thillet J, Hoang-Xuan K (2004) Molecular changes in

gliomas. Curr Opin Oncol 16:607–613

36. Ishihara K, Tsutsumi K, Kawane S, Nakajima M, Kasaoka T

(2003) The receptor for advanced glycation end-products

(RAGE) directly binds to ERK by a D-domain-like docking site.

FEBS Lett 550:107–113

37. Giese A, Loo MA, Tran N, Haskett D, Coons SW, Berens ME

(1996) Dichotomy of astrocytoma migration and proliferation. Int

J Cancer 67:275–282

38. Fages C, Nolo R, Huttunen HJ, Eskelinen E, Rauvala H (2000)

Regulation of cell migration by amphoterin. J Cell Sci 113:

611–620

39. Qiu RG, Chen J, Kirn D, McCormick F, Symons M (1995) An

essential role for Rac in Ras transformation. Nature 374:457–459

40. Chan AY, Coniglio SJ, Chuang YY, Michaelson D, Knaus UG,

Philips MR, Symons M (2005) Roles of the Rac1 and Rac3

GTPases in human tumor cell invasion. Oncogene 24:7821–7829

41. Chuang YY, Tran NL, Rusk N, Nakada M, Berens ME, Symons

M (2004) Role of synaptojanin 2 in glioma cell migration and

invasion. Cancer Res 64:8271–8275

42. Salhia B, Rutten F, Nakada M, Beaudry C, Berens M, Kwan A,

Rutka JT (2005) Inhibition of Rho-kinase affects astrocytoma

morphology, motility, and invasion through activation of Rac1.

Cancer Res 65:8792–8800

43. Passalacqua M, Zicca A, Sparatore B, Patrone M, Melloni E,

Pontremoli S (1997) Secretion and binding of HMG1 protein to

the external surface of the membrane are required for murine

erythroleukemia cell differentiation. FEBS Lett 400:275–279

44. Kuniyasu H, Yano S, Sasaki T, Sasahira T, Sone S, Ohmori H

(2005) Colon cancer cell-derived high mobility group 1/ampho-

terin induces growth inhibition and apoptosis in macrophages.

Am J Pathol 166:751–760

45. Wahamaa H, Vallerskog T, Qin S, Lunderius C, LaRosa G,

Andersson U, Harris HE (2007) HMGB1-secreting capacity of

multiple cell lineages revealed by a novel HMGB1 ELISPOT

assay. J Leukoc Biol 81:129–136

46. Brezniceanu ML, Volp K, Bosser S, Solbach C, Lichter P, Joos S,

Zornig M (2003) HMGB1 inhibits cell death in yeast and mam-

malian cells and is abundantly expressed in human breast

carcinoma. FASEB J 17:1295–1297

47. Ditsworth D, Zong WX, Thompson CB (2007) Activation of

poly(ADP)-ribose polymerase (PARP-1) induces release of the

pro-inflammatory mediator HMGB1 from the nucleus. J Biol

Chem 282:17845–17854

48. Wharton B, McNelis U, Bell HS, Whittle IR (2000) Expression of

poly(ADP-ribose) polymerase and distribution of poly(ADP-

ribosyl)ation in glioblastoma and in a glioma multicellular

tumour spheroid model. Neuropathol Appl Neurobiol 26:528–535

49. Raza SM, Lang FF, Aggarwal BB, Fuller GN, Wildrick DM,

Sawaya R (2002) Necrosis and glioblastoma: a friend or a foe? A

review and a hypothesis. Neurosurgery 51:2–12

J Neurooncol (2008) 87:23–33 33

123