Stimulation of excitatory amino acid release from adult mousebrain glia subcellular particles by high mobility group box 1protein

Marco Pedrazzi,*,1 Luca Raiteri,*,1 Giambattista Bonanno,* Mauro Patrone,� Sabina Ledda,*Mario Passalacqua,* Marco Milanese,* Edon Melloni,* Maurizio Raiteri,* Sandro Pontremoli*and Bianca Sparatore*

*Department of Experimental Medicine and Center of Excellence for Biomedical Research, University of Genoa, Genoa, Italy

�Department of Environmental and Life Sciences, University of Eastern Piedmont ‘Amedeo Avogadro’, Alessandria, Italy

Abstract

The multifunctional protein high mobility group box 1 (HMGB1)

is expressed in hippocampus and cerebellum of adult mouse

brain. Our aim was to determine whether HMGB1 affects

glutamatergic transmission by monitoring neurotransmitter

release from glial (gliosomes) and neuronal (synaptosomes)

re-sealed subcellular particles isolated from cerebellum and

hippocampus. HMGB1 induced release of the glutamate

analogue [3H]D-aspartate form gliosomes in a concentration-

dependent manner, whereas nerve terminals were insensitive

to the protein. The HMGB1-evoked release of [3H]D-aspartate

was independent of modifications of cytosolic Ca2+ , but it was

blocked by DL-threo-b-benzyloxyaspartate (DL-TBOA), an

inhibitor of glutamate transporters. HMGB1 also stimulated

the release of endogenous glutamate in a Ca2+-independent

and DL-TBOA-sensitive manner. These findings suggest the

involvement of carrier-mediated release. Moreover, dihydro-

kainic acid, a selective inhibitor of glutamate transporter 1

(GLT1), does not block the effect of HMGB1, indicating a role

for the glial glutamate-aspartate transporter (GLAST) subtype

in this response. We also demonstrate that HMGB1/glial

particles association is promoted by Ca2+. Furthermore, al-

though HMGB1 can physically interact with GLAST and the

receptor for advanced glycation end products (RAGE), only its

binding with RAGE is promoted by Ca2+. These results sug-

gest that the HMGB1 cytokine could act as a modulator of

glutamate homeostasis in adult mammal brain.

Keywords: glial glutamate-aspartate transporter; gliosomes;

glutamate release; high mobility group box 1; receptor for

advanced glycation end products; synaptosomes.

J. Neurochem. (2006) 99, 827–838.

The high mobility group box 1 (HMGB1) protein is achromatin binding-factor highly conserved in mammals(Bustin 1999). This protein is widely expressed in thedeveloping nervous system and in transformed cells of nerveorigin, including neuroblastomas and gliomas (Taguchi et al.2000; Sajithlal et al. 2002; Guazzi et al. 2003). It has beenshown that, despite the lack of a classical signal peptide forprotein secretion, HMGB1 can be actively exported byseveral cell types (Muller et al. 2004). The release ofHMGB1 by pituicytes, promoted by pro-inflammatorycytokines, suggests its possible role in the regulation ofneuroendocrine and immune responses to infection or injury(Wang et al. 1999). Extracellular HMGB1 has also beenproposed as a modulator of nervous tissue histogenesis, inthat it shows high affinity binding properties to the

Received May 15, 2006; revised manuscript received July 4, 2006;accepted July 9, 2006.Address correspondence and reprint requests to Giambattista

Bonanno, Department of Experimental Medicine, Pharmacology andToxicology Section, Viale Cembrano 4, 16148 Genoa, Italy.E-mail: bonanno@pharmatox.unige.it1These two authors contributed equally to this work.Abbreviations used: AEBSF, 4-(2-aminoethyl)benzenesulfonyl fluor-

ide hydrochloride; [3H]D-Asp, [3H]D-aspartate; BAPTA-AM, bis(2-aminophenyl)ethyleneglycol-N,N,N¢,N¢-tetraacetic acid, tetraacetoxy-methyl ester; DHK, dihydrokainic acid; GFAP, glial fibrillary acidicprotein; GLAST, glutamate-aspartate transporter; GLT1, glutamatetransporter 1; HMGB1, high mobility group box 1; MAP-2, microtubule-associated protein-2; RAGE, receptor for advanced glycation endproducts; DL-TBOA, DL-threo-b-benzyloxyaspartate.

Journal of Neurochemistry, 2006, 99, 827–838 doi:10.1111/j.1471-4159.2006.04120.x

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 99, 827–838 827

chondroitin sulphate proteoglycans neurocan and phosph-acan and activates phosphacan/contactin interaction (Milevet al. 1998). In addition, in developing rat brain HMGB1colocalises with the receptor for advanced glycation endproducts (RAGE; Hori et al. 1995). Engagement of thismultiligand receptor by HMGB1 induces filopodia andneurite extension in neuroblastoma, mediated by membersof the Rho family (Huttunen et al. 1999), and enhancesproliferation and metastatic invasiveness of glioma cellsthrough cytoskeletal reorganisation and activation of cellmotility, which involve mitogen-activated protein kinasescascades (Taguchi et al. 2000).

Extracellular HMGB1 shows characteristics typical of pro-inflammatory cytokines and behaves as a mediator of theinnate immune system (Messmer et al. 2004; Semino et al.2005). Recently, pro-inflammatory effects of HMGB1,identified in specific structures of the peripheral and thecentral nervous system, have supported the concept that thisprotein may operate through direct activation of signallingcascades, commonly utilised by classic pro-inflammatorycytokines, or indirectly by inducing the release of thesecytokines (Agnello et al. 2002; O’Connor et al. 2003; Ronget al. 2004). HMGB1 has been also proposed as a newpossible target for therapeutic strategies in Alzheimer’sdisease, as this protein factor has been detected in associationwith senile plaques and found to be able to enhanceaccumulation of extracellular amyloid-beta peptides byinhibiting microglial phagocytosis (Takata et al. 2004).Taken together, these findings indicate that extracellularHMGB1 can behave as a neurotrophic factor, as a pro-inflammatory cytokine or as an enhancer of neurodegener-ation, depending on the developmental stage of the nervoussystem, the concomitant signal activity of other effectors andthe occurrence of necrotic events that accumulate HMGB1and HMGB1-binding proteins in the extracellular matrix.

Although HMGB1 is widely expressed in the developingcentral nervous system, the adult brain maintains high levelsof HMGB1 only in specific areas, including the hippocampusand the cerebellum (Guazzi et al. 2003). To evaluate whetherextracellular HMGB1 could modulate functional activities inthe central nervous system of adult mouse, we monitored therelease of glutamate from purified axon terminals (synapto-somes; Gray and Whittaker 1962) and glial plasmalemmalvesicles, here referred as to gliosomes (Nakamura et al.1993).

Materials and methods

Animals

Adult male mice (Swiss, 7–8 weeks old, Charles River Laborator-

ies) were used. Animals were housed at constant temperature

(22 ± 1�C) and relative humidity (50%) under a regular light–dark

schedule (lights on at 07.00 h and off at 19.00 h). Food and water

were freely available. The experiments were carried out in

accordance with the European Community Council Directive of

24 November 1986 (86/609/EEC). All efforts were made to

minimise animal suffering and to use only the number of animals

necessary to produce reliable results.

Preparation of synaptosomes and gliosomes

Animals were sacrificed by cervical dislocation and the cerebella

and hippocampi quickly dissected. Purified glial re-sealed particles

(gliosomes) and synaptosomes were prepared essentially according

to Nakamura and collaborators (1994), with some modifications.

The tissue was homogenised in 10 volumes of 10 mM Tris/HCl

pH 7.4 containing 0.32 M sucrose, using a glass-Teflon tissue

grinder (clearance 0.25 mm; 12 up–down strokes in about 1 min at

900 rpm/min or 24 up–down strokes in about 2 min at 90 rpm/min

for hippocampus or cerebellum, respectively). The homogenate was

centrifuged (5 min at 4�C and 1000 g) to remove nuclei and debris

and the supernatant stratified on a discontinuous Percoll� gradient

(2, 6, 10 and 20% (v/v) in Tris-buffered sucrose) and centrifuged for

5 min at 4�C and 33 500 g. The layers between 2% and 6% (v/v)

Percoll (gliosomal fraction) and between 10% and 20% (v/v) Percoll

(synaptosomal fraction) were collected and washed and diluted for

biochemical studies in the following medium: 140 mM NaCl, 3 mM

KCl, 1.2 mM MgCl2, 1.2 mM NaH2PO4, 10 mM HEPES, 10 mM

glucose; pH 7.4 (medium A). For release experiments the gliosomal

and synaptosomal fractions were washed and diluted in the

following medium: 125 mM NaCl, 3 mM KCl, 1.2 mM MgSO4,

1.2 mM CaCl2, 1 mM NaH2PO4, 22 mM NaHCO3, 10 mM glucose

(aeration with 95% O2 and 5% CO2); pH 7.2–7.4.

Confocal microscopy

Gliosomes and synaptosomes (75 lg protein) were placed onto

coverslips pretreated with poly-L-ornithine. The preparations were

fixed with 2% (w/v) paraformaldehyde (15 min at 4�C) and

permeabilised with 0.05% (v/v) Triton X-100 (5 min at 4�C).Rabbit anti-glial fibrillary acidic protein (GFAP; 1 : 100) and mouse

antimicrotubule-associated protein-2 (MAP-2; 1 : 200, 30 min at

20�C) were used as primary antibodies; goat anti-rabbit Texas Red

or donkey anti-mouse Alexa 488-labelled (1 : 200, 30 min at 20�C)were used as secondary antibodies. Images were collected by

confocal microscopy using a krypton/argon laser MRC1034

instrument (Bio-Rad, Segrate, Milan, Italy) attached to a Nikon

Diaphot 200 inverted microscope (Nikon Inc., Tokyo, Japan), using

a 60· Plan Apo oil-immersion objective with numerical aperture

1.4. The excitation/emission wavelengths were 488/522 and 567/

605 for Alexa 488- and Texas Red-labelled antibodies, respectively.

Release experiments

Synaptosomes and gliosomes were incubated for 15 min at 37�C in

the absence or presence of 0.1 lM [3H]D-Asp (experiments of

endogenous glutamate release). In some experiments, gliosomes

were incubated 30 min (15 min before and during labelling with

[3H]D-Asp) in the presence of 100 lM bis(2-aminophenyl)ethylene-

glycol-N,N,N¢,N¢-tetraacetic acid, tetraacetoxymethyl ester (BAP-

TA-AM). Aliquots (6 lg protein) of the suspensions were distributedon microporous filters placed at the bottom of a set of parallel

superfusion chambers maintained at 37�C (Superfusion System,

Ugo Basile, Comerio, Varese, Italy). Superfusion was started at a rate

828 M. Pedrazzi et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 99, 827–838� 2006 The Authors

of 0.5 mL/min with the release experiment medium specified above

(Raiteri et al. 1974; Raiteri and Raiteri 2000). After 33 min, when

the system had been allowed to equilibrate, eight (HMGB1

experiments) or five (ATP experiments) 3-min samples were

collected. Synaptosomes and gliosomes were exposed to HMGB1

or ATP at the end of the second sample collected (t ¼ 39 min) until

the end of experiment. DL-threo-b-benzyloxyaspartic acid (DL-

TBOA) or dihydrokainic acid (DHK) was introduced at t ¼30 min. When appropriate, Ca2+ was omitted at t ¼ 20 min. The

Ca2+-free medium contained 500 lM EGTA. Radioactivity was

determined in samples collected and superfused filtres by liquid

scintillation counting. The tritium present in the superfusate

samples was expressed as a percentage of the radioactivity content

of the particles at the start of the respective collection period

(fractional rate · 100). Endogenous glutamate was expressed as

nmol/mg of gliosomal protein. HMGB1-evoked [3H]D-Asp or

endogenous glutamate overflow was calculated by subtracting the

appropriate basal efflux from the neurotransmitter content in the

stimulated release samples. The concentration-response curves

were fitted to the experimental data using the following four

parameters logistic equation, provided by the software SIGMA PLOT

version 8.0: y ¼ a + {(b–a)/[1 + (10c/10x)d]}, where a is the

minimum and b the maximum value, c is the EC50, and d is the

slope of the curve.

Endogenous glutamate determination

Endogenous glutamate was measured by high performance liquid

chromatography analysis following precolumn derivatisation with

o-phthalaldehyde and separation on a C18 reverse-phase chroma-

tographic column (10 · 4.6 mm, 3 lm; at 30�C; Chrompack,

Middleburg, the Netherlands) coupled with fluorometric detection

(excitation wavelength 350 nm; emission wavelength 450 nm;

Raiteri et al. 2000). The following buffers were utilised: solventA, 0.1 M sodium acetate (pH 5.8)/methanol 80 : 20; solvent B, 0.1 M

sodium acetate (pH 5.8)/methanol, 20 : 80; solvent C, sodium

acetate (pH 6.0)/methanol, 80 : 20. The gradient program was as

follows: 100% C for 4 min from the initiation of the program; 90%

A and 10% B in 1 min; 42% A and 58% B in 14 min; 100% B in

1 min; isocratic flow 2 min; 100% C in 3 min; flow rate 0.9 mL/

min. Homoserine was used as an internal standard.

Purification and radiolabelling of HMGB1

Eukaryotic recombinant HMGB1 protein was produced by means of

the baculovirus system and purified as described (Sparatore et al.1996). The amount of lipopolysaccharide in HMGB1 preparations

was assayed by the Pyrotell test and was 2.5–3 pg lipopolysaccha-

ride/lg HMGB1. [125I]-labelling of the purified protein was carried

out as reported previously (Passalacqua et al. 1998) and its specific

radioactivity was 1.75 · 105 cpm/pmol.

Isolation of detergent-soluble fraction from gliosomes and

co-immunoprecipitations

Gliosomes (500 lg total protein), obtained as described above, were

washed twice in medium A and then collected by centrifugation for

15 min at 4�C and 14 000 g. The pellet was diluted and lysed in

1 mL of ice-cold 20 mM Tris/HCl pH 7.4, containing 10 mM NaCl,

100 lg/mL leupeptin, 10 lg/mL aprotinin and 2 mM 4-(2-amino-

ethyl)benzenesulfonyl fluoride hydrochloride (AEBSF). After

15 min at 4�C the lysate was centrifuged for 15 min at 4�C and

200 000 g, the pellet (cell membranes) was diluted in 1 mL of

20 mM Tris/HCl pH 7.4, containing 140 mM NaCl, the protease

inhibitors indicated above and 1.2 mM Ca2+ or, where indicated,

5 mM EDTA. Cell membranes were finally solubilised by adding

Triton X-100 (0.5% v/v final concentration). After 15 min at 4�C the

sample was centrifuged for 15 min at 4�C and 200 000 g, the

supernatant (detergent-soluble membrane fraction) was collected

and protein concentration measured using the BCA Protein Assay

Kit. Immunoprecipitations were carried out using 50 lg of

detergent-soluble membrane proteins incubated with 10 nM

HMGB1 in 0.5 mL of immunoprecipitation buffer. After 10 min

at 4�C, samples were subjected to the immunoprecipitation

procedure described elsewhere (Sparatore et al. 2005), using 2 lgof mouse anti-HMGB1 antibody (Sparatore et al. 2005) or 2 lg of

antiglutamate-aspartate transporter (GLAST) antibody in the pres-

ence of 1.2 mM Ca2+ or 5 mM EDTA.

Evaluation of RAGE/HMGB1 interaction

The gliosome detergent-soluble fraction, obtained as specified

above, was diluted in 0.5 mL of 20 mM Tris/HCl pH 7.4, containing

140 mM NaCl, the protease inhibitors indicated above, 0.5% (v/v)

Triton X-100 and 1.2 mM Ca2+ or, when indicated, 5 mM EDTA in

the absence of Ca2+ and incubated with 30 l1 CNBr-activated

Sepharose� 4B, to which purified HMGB1 had been covalently

bound (2.6 mg HMGB1/mL resin), according to the manufacturer’s

instructions. After 2 h at 4�C the resin was subjected to 5 · 0.2 mL

washes with the specific buffer, followed by 5 · 0.2 mL washes

with the same buffer in the absence of Triton X-100. The adsorbed

proteins were eluted by heating the resin 5 min at 95�C in sodium

dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE)

loading sample buffer.

Evaluation of particles/HMGB1 interaction

[125I]-labelled HMGB1 (125 ng) was incubated with 50 lg of

gliosomes or synaptosomes in 0.5 mL of medium A containing

0.1% (w/v) bovine serum albumin and 1.2 mM Ca2+, 12 mM Ca2+,

or 500 lM EGTA in the absence of Ca2+ for 10 min at 4�C, in the

absence or in the presence of 100-fold molar excess of unlabelled

HMGB1. Particles were collected by centrifugation for 5 min at 4�Cand 14 000 g and washed with 1 mL of binding medium; bound

radioactivity was measured with a c-counter (Cobra II, Packard-

Perkin Elmer, Monza, Italy).

Immunoblotting

Proteins were separated by SDS)10% PAGE and transferred onto

nitrocellulose membranes. Membranes were blocked in 20 mM

sodium phosphate buffer, pH 7.4, containing 140 mM NaCl, 5%

(w/v) nonfat dry milk, 0.1% (v/v) Tween-20 and incubated with one

of the following primary antibodies (60 min at 20�C): guinea pig

anti-GLAST (1 : 5000), guinea pig anti-glutamate transporter 1

(anti-GLT1; 1 : 10 000), mouse anti-RAGE (1: 2000), mouse anti-

HMGB1 (1 : 4000), mouse anti-GFAP (1 : 1000), mouse anti-b-tubulin III (1 : 1000). Immunoreactivities were then evaluated with

peroxidase-conjugated secondary antibodies (60 min at 20�C): anti-guinea pig (1 : 4000), anti-mouse (1 : 3000), followed by develop-

ment with an ECL� (enhanced chemiluminescence) Plus Western

blotting detection system. Quantification of immunoreactive bands

HMGB1-promoted glutamate release in gliosomes 829

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 99, 827–838

was carried out with a dual-wavelength flying-spot scanner

(Shimadzu Corporation, Tokyo, Japan).

Statistical analysis

The two-tailed Student’s t-test was used for statistical comparison of

the data.

Materials

[3H]D-Aspartate ([3H]D-Asp) specific activity 16.3 Ci/mmol),

horseradish peroxidase-conjugated donkey anti-guinea pig secon-

dary antibody, sheep anti-rabbit secondary antibody and the ECL

Plus Western blotting detection system were purchased from

Amersham Biosciences (Cologno Monzese, Milan, Italy). Horse-

radish peroxidase-conjugated horse anti-mouse secondary antibody

was purchased from Cell Signaling Technology (Celbio, Pero,

Milan, Italy). The Alexa 488-conjugated donkey anti-mouse

antibody was obtained from Molecular Probes (Invitrogen,

S. Giuliano Milanese, Milan, Italy). Mouse anti-MAP-2, guinea

pig anti-GLAST, guinea pig anti-GLT1 and mouse anti-RAGE

antibodies were acquired from Chemicon International (Prodotti

Gianni, Milan, Italy). Rabbit anti-GFAP, mouse anti-GFAP and

mouse anti-b-tubulin III antibodies, aprotinin, Percoll, BAPTA-

AM, ATP and Triton X-100 were obtained from Sigma Aldrich

(Milan, Italy). The goat anti-rabbit Texas Red-conjugated secon-

dary antibody and AEBSF were obtained from Calbiochem (VWR,

Milan, Italy). DL-TBOA and DHK were from Tocris Cookson

(Bristol, UK). The BCA Protein Assay Kit was purchased from

Pierce (Celbio, Pero, Milan, Italy).

Results

Gliosomes purified from adult rat cerebral cortex containedvery low amounts of synaptosomes in a recent thoroughcharacterisation (Stigliani et al. 2006). To ascertain thelevel of purification of gliosomal- and synaptosomal-enriched fractions isolated in this study from adult mousebrain, we analysed the distribution of glial and neuronalmarker proteins by confocal microscopy and westernblotting. In confocal microscopy experiments, each purifiedpreparation, isolated from adult mouse, was stained todetect the proportion of particles expressing the glialspecific marker GFAP or the neuronal marker MAP-2. Asshown in Fig. 1a, both cerebellar and hippocampal glio-somes contained only a minority of particles of neuronalorigin and synaptosomes were scantly contaminated byglial-derived particles. No staining was detectable whengliosome and synaptosome preparations were subjected toincubation with secondary antibodies alone (not shown).The counting of red and green pixels in diverse imagesobtained from three different synaptosome and gliosomepreparations indicated that no more than 15% GFAPlabelling was present in the synaptosomal fraction and,conversely, that a maximum of 20% MAP-2 labelling waspresent in gliosomal preparations, both in hippocampusand in cerebellum.

Cerebellum

Glio

som

esS

ynap

toso

mes

Hippocampus(a)

(b)

MAP-2 GFAP

GFAP

MAP-2

MAP-2MAP-2

GFAP

GFAP

Synaptoso

mes

GFAP

Glioso

mes

Synaptoso

mes

Glioso

mes

β-tubulin III

Hippocampus Cerebellum

Fig. 1 Distribution of protein markers in

hippocampal and cerebellar synaptosome

and gliosome preparations. (a) Synapto-

somes and gliosomes were placed onto

coverslips fixed, permeabilised and incu-

bated with the primary and secondary anti-

bodies as specified in the Materials and

methods. Microtubule-associated protein-2

(MAP-2, green fluorescence) and glial

fibrillary acidic protein (GFAP, red fluo-

rescence) were analysed by confocal micr-

oscopy. Images are representative of three

separated particle preparations. Scale bar,

40 mm. (b) Western blotting of GFAP and

b-tubulin III was performed using 10 lg

whole synaptosomes or gliosomes from

mouse hippocampus or cerebellum. A rep-

resentative experiment (of three) is shown.

830 M. Pedrazzi et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 99, 827–838� 2006 The Authors

Due to the high mass of MAP-2 (about 300 kDa) we couldnot study the expression of this protein by western blotting.Therefore, we used b-tubulin III as a synaptosomal marker.As shown in Fig. 1b, b-tubulin III appeared to be representedin much greater amounts in the synaptosomal fraction than inpurified gliosomes; conversely, the astrocyte marker GFAPcould be detected in gliosomes and much less so insynaptosomes. These results are in agreement with previousreports demonstrating that gliosomes and synaptosomes canbe largely, although not completely, separated (Matsui andJahr 2004; Raiteri et al. 2005; Stigliani et al. 2006).However, we obtained a level of glial and neuronalenrichment sufficient to allow a direct comparison betweenglutamatergic responses depending on mechanisms specific-ally operating in nerve endings or glial particles.

Effect of exogenous HMGB1 on the release of [3H]D-Asp

from synaptosomes and gliosomes

To make a direct comparison of the ability of HMGB1 tomodulate glutamate release from nerve endings and glialparticles belonging to the same brain area, synaptosomes andgliosomes were prepared from adult mouse hippocampus orcerebellum. The preparations were labelled with the stableglutamate analogue [3H]D-Asp and exposed to HMGB1 in

superfusion. As shown in Fig. 2, no increase in thespontaneous efflux of the radioactive amino acid wasobserved in hippocampal (Fig. 2a, left) or cerebellar(Fig. 2b, left) synaptosomes exposed to 3 nM HMGB1. Bycontrast, HMGB1 promoted a concentration-dependentrelease of [3H]D-Asp from hippocampal and cerebellargliosomes, when administered at 1 and 3 nM (Fig. 2a,b,right). To better characterise the different sensitivity ofgliosomes and synaptosomes to HMGB1, we constructedconcentration-response curves. HMGB1 stimulated therelease of [3H]D-Asp in a concentration-dependent mannerfrom hippocampal (Fig. 3a) and from cerebellar (Fig. 3b)gliosomes. The two concentration–response curves appearedsimilar: the maximal potentiation of release was obtainedaround 3 nM HMGB1 and the EC50 amounted to0.57 ± 0.043 nM and to 0.32 ± 0.041 nM for hippocampaland cerebellar gliosomes, respectively. By contrast, nosignificant increase of [3H]D-Asp efflux was observed insynaptosomes under the same experimental conditions.

These findings indicate that (i) nanomolar concentrationsof extracellular HMGB1 are sufficient to elicit glutamaterelease from gliosomes and (ii) this signalling activity ofHMGB1 is specifically directed to glial and not to neuronalpreparations from the same regions of adult mouse brain.

42 45 48 54 5763 93

6

7

8

9

10

11

12

13

14 3 nM HMGB1Control

51

[3 H

]D-A

sp r

elea

se(f

ract

iona

l rat

e x

100)

[3 H]-

D-A

sp r

elea

se

(fr

actio

nal r

ate

x 10

0)

42 45 48 51 54 5763 93

Synaptosomes

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0 3 nM HMGB1Control

42 45 48 54 573936

6

7

8

9

10

11

12

13

14 3 nM HMGB11 nM HMGB1Control

51

42 45 48 51 54 573936

Gliosomes

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

3 nM HMGB11 nM HMGB1 Control

Time (min)

Hippocampus

Synaptosomes Gliosomes

Time (min)

Cerebellum

(a)

(b)

Fig. 2 Time course of the HMGB1-evoked

release of [3H]D-Asp from mouse hippo-

campal (a) and cerebellar (b) synapto-

somes and gliosomes. Purified particles

were labelled with the radioactive tracer.

HMGB1 was added to the superfusion

medium at the end of the second sample

collected (see arrow). At the indicated

times, 1.5-mL samples of superfusate

medium were counted for radioactivity.

Results are expressed as fractional

rate · 100 (percentage of the total radio-

activity content of gliosomes or synapto-

somes at the start of the respective

collection period). The data presented are

means ± SEM of four experiments in tripli-

cate.

HMGB1-promoted glutamate release in gliosomes 831

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 99, 827–838

Effectors of the release of [3H]D-Asp promoted by

HMGB1 from gliosomes

The release of glutamate can be mediated by Ca2+-dependentand Ca2+-independent mechanisms (Attwell et al. 1993; Leviand Raiteri 1993; Berridge 1998). To determine whether theefflux [3H]D-Asp promoted by HMGB1 involved vesicularexocytosis, which is known to require the entry of Ca2+

through voltage-sensitive calcium channels or mobilisationof Ca2+ from internal stores to the cytosol (Berridge 1998),we assayed the effect of HMGB1 either by omitting Ca2+

from the extracellular medium or by using gliosomespretreated with BAPTA-AM, a cytosolic Ca2+ chelator. Asshown in Fig. 4, omission of external Ca2+, in the presenceof 500 lM EGTA, largely prevented the HMGB1 effect,suggesting the occurrence of exocytotic [3H]D-Asp release.Similar results were obtained using a Ca2+-free mediumsupplemented with 8.8 mM MgCl2 (not shown). Surprisingly,in the presence of external Ca2+, the HMGB1-evoked [3H]D-

Asp release did not prove to be significantly modified ingliosomes pretreated with BAPTA-AM, thus minimising theexocytotic component in the process. The lack of effectobserved with BAPTA-AM does not appear to be due to theincapability of the Ca2+ chelator to remove cytosolic Ca2+ inthe presence of 1.2 mM external Ca2+. In fact, in the course ofexperiments analysing the effect of BAPTA-AM on the Ca2+-dependent release of [3H]D-Asp induced by ATP, which hasbeen shown previously to occur in cerebral cortex gliosomes(Stigliani et al. 2006), exposure of hippocampal gliosomes to3 mM ATP resulted in a stimulation of the radioactive tracerrelease (percent overflow 6.82 ± 0.47, n ¼ 4); an effectreduced to about 60% by BAPTA-AM in the presence ofexternal Ca2+ (percent overflow 4.25 ± 0.35, n ¼ 4,p < 0.05).

The discrepancy between external Ca2+ dependency andBAPTA-AM independency of the HMGB1-evoked [3H]D-Asp release prompted us to check the effect of thenontransportable blocker of glutamate uptake DL-TBOA onthe amino acid release elicited by HMGB1. As shown inFig. 4, DL-TBOA (10 lM) eliminated the efflux of [3H]D-Asppromoted by 3 nM HMGB1, suggesting that this response isdue to a mechanism mediated by reversal of glutamatetransporters (Attwell et al. 1993; Levi and Raiteri 1993).Interestingly, DHK (50 lM), a selective inhibitor of GLT1glutamate transporter (Arriza et al. 1994), did not signifi-cantly modify the HMGB1-induced [3H]D-Asp overflow.

-10123456789

0.1

[3 H]-

D-A

sp r

elea

se (

% o

verf

low

) [3 H

]-D

-Asp

rel

ease

(%

ove

rflo

w)

1 100.01

SynaptosomesGliosomes

-10123456789

0.1

HMGB1 concentration (nM)

HMGB1 concentration (nM)1 100.01

SynaptosomesGliosomes

Hippocampus

Cerebellum

(a)

(b)

Fig. 3 Concentration–response curve of the HMGB1-evoked

[3H]D-Asp release from mouse hippocampal (a) or cerebellar (b) syn-

aptosomes and gliosomes. Purified particles were labelled with the

radioactive tracer and exposed in superfusion to various concentra-

tions of HMGB1. Samples of superfusate were collected and counted

for radioactivity. Results are expressed as percentage overflow

evoked by HMGB1. The data presented are means ± SEM of 4–15

experiments in triplicate.

[3 H]-

D-A

sp r

elea

se (

% o

verf

low

)

-

0

1

2

3

4

5

3 nM HMGB1

-- - -

+

- - -

*

--

- - - + --

+--

- -

*

- - -

-+-

-

--

-

+ + + +++ + +Ca2+-free + 500 µM EGTA

100 µM BAPTA-AM

10 µM DL-TBOA

50 µM DHK

Hippocampus gliosomes

Fig. 4 Effects of Ca2+ omission, of the Ca2+ chelator BAPTA-AM and

of the nontransportable glutamate carrier blockers DL-TBOA or DHK

on the HMGB1-evoked [3H]D-Asp release from hippocampal glio-

somes. HMGB1 was present in superfusion from the end of the sec-

ond fraction collected. Ca2+-free medium, containing 500 lM EGTA,

was introduced 19 min before HMGB1; DL-TBOA or DHK 9 min before

HMGB1. BAPTA-AM was incubated with synaptosomes 30 min at

37�C before and during labelling. Results are expressed as percent-

age overflow evoked by HMGB1. The data are means ± SEM of four to

seven experiments in triplicate. *p < 0.001 vs. the respective control

value representing the effect of HMGB1 in the absence of drugs (two-

tailed Student’s t-test).

832 M. Pedrazzi et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 99, 827–838� 2006 The Authors

Finally, a set of experiments have been carried out tomeasure endogenous glutamate release in hippocampus andthe effect of HMGB1. As shown in Fig. 5, HMGB1 alsostimulated the spontaneous release of endogenous glutamatefrom hippocampal gliosomes; the effect showing a time-course basically superimposable on that seen with [3H]D-Asp. The enhancement due the cytokine was stronglyreduced (about 70%) in the absence of external Ca2+ andabolished in the presence of the glutamate transport blockerDL-TBOA, as in the case of the radioactive tracer.

In sum, the results with [3H]D-Asp and endogenousglutamate suggest that the release from hippocampal glio-somes exposed to exogenous HMGB1 involves a carrier-mediated process and concurrently is dependent on externalCa2+. However, as carrier-mediated release of neurotrans-mitters does not depend on external Ca2+ (Attwell et al.1993; Levi and Raiteri 1993), we further explored the originof this discrepancy.

Binding properties of HMGB1 to gliosomes and

synaptosomes

To determine whether the lack of HMGB1-induced release of[3H]D-Asp, observed in the absence of external Ca2+ , was aconsequence of a real Ca2+ requirement for the amino acidexport or depended on altered gliosome/HMGB1 interactionin this condition, we measured the binding of [125I]HMGB1to gliosomes in the absence or in the presence of externalCa2+. As shown in Fig. 6, the labelled protein bound togliosomal particles isolated from hippocampal and cerebellarregions, but the amount recovered in association with these

particles was significantly reduced when Ca2+ was omittedfrom the binding medium. The results reported in Fig. 6suggest that the ion promotes the interaction of exogenousHMGB1 with the gliosome surface, even at micromolarconcentrations. In contrast, synaptosomes, purified from thesame brain areas, bound very low amounts of the cytokineboth in the presence (Fig. 6) and in the absence of Ca2+ (datanot shown), suggesting that glial particles preferentiallycontain Ca2+-activated binding site(s) for HMGB1, incontrast to nerve endings.

Two glutamate transporters have been reported to beexpressed in glial cells, namely GLAST and GLT1 (Schmittet al. 1997; Sims and Robinson 1999). As shown in Fig. 7a,comparison of the amounts of GLAST and GLT1 detected byimmunoblotting in gliosome and synaptosome preparationsconfirmed that these transporters are exclusively (GLAST) orpreferentially (GLT1) represented in the gliosome fraction.The quantification of immunoreactive signals, obtained withdifferent preparations, revealed 4.5-fold and 1.8-fold higheramounts of GLAST and GLT1 in gliosomes than insynaptosomes, respectively. We next sought to determinewhether glutamate transporters could interact directly withHMGB1 and whether Ca2+ influenced these protein–proteinbindings. Detergent-soluble membrane fractions, extractedfrom hippocampal gliosomes in a Ca2+-containing (lane 1) or

Hippocampal gliosomes

1.0

1.2

1.4

1.6

1.8

2.03 nM HMGB1

3 nM HMGB1 in Ca2+-free medium

Control

End

ogen

ous

glut

amat

e re

leas

e(n

mol

/mg

prot

ein)

3 nM HMGB1 +10 µM DL-TBOA

time (min)36 39 42 45 48 51 54 57

Fig. 5 Time course of the HMGB1-evoked release of endogenous

glutamate from mouse hippocampal gliosomes and effects of Ca2+

omission or DL-TBOA. HMGB1 was added to the superfusion medium

at the end of the second sample collected (see arrow). Ca2+-free

medium, containing 500 lM EGTA, was introduced 19 min and DL-

TBOA 9 min before HMGB1. At the indicated times, samples were

collected and counted for endogenous neurotransmitter content.

Results are expressed as nmol/mg of gliosomal protein. The data

presented are means ± SEM of three experiments in triplicate.

HM

GB

1 sp

ecifi

c bi

ndin

g(p

mol

/mg

prot

ein)

0

1

2

3

4

5

6

* *

HippocampusCerebellum

Gliosomes Synaptosomes

1.2 mM Ca2+ 12 µM Ca2+ Ca2+-free500 µM EGTA

Fig. 6 Binding of [125I]-HMGB1 to gliosomes and synaptosomes from

mouse cerebellum or hippocampus. The specific binding of the radi-

olabelled protein to gliosomes or synaptosomes was evaluated in the

absence or presence of a 100-fold molar excess of unlabelled HMGB1

in a standard medium containing 1.2 mM CaCl2 or 12 lM CaCl2 or in a

Ca2+-free medium containing 500 lM EGTA as specified in Materials

and methods. In the absence of CaCl2 the amount of HMGB1 recov-

ered bound to synaptosomes was not significantly different from that

measured in the presence of the ion (data not shown). Values repre-

sent the means ± SEM of three separate experiments. *p < 0.005 vs.

the respective control value representing the binding of HMGB1 in the

presence of Ca2+ (two-tailed Student’s t-test).

HMGB1-promoted glutamate release in gliosomes 833

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 99, 827–838

a Ca2+-free (lane 2) medium, revealed the presence of similaramounts of GLAST and GLT1 and were used to performHMGB1 co-immunoprecipitation experiments (Fig. 7b).GLAST co-immunoprecipitated with HMGB1 and theamounts of GLAST recovered in the HMGB1 co-immuno-precipitated material in the presence (lane 1) or the absence(lane 3) of Ca2+ were not significantly different (Fig. 7c). Bycontrast, GLT1 was detectable only in the pool of proteinsnot co-immunoprecipitated with HMGB1 (lanes 2 and 4). Wealso examined whether HMGB1 could be co-immunopre-

cipitated using an anti-GLAST antibody. Also in thiscondition we detected co-immunoprecipitation of HMGB1and GLAST (Fig. 7c, bottom, lanes 1 and 3), confirming thephysical interaction between the two proteins. These findingsindicate that HMGB1 can associate with GLAST, but notwith GLT1, and that this protein–protein interaction is notaffected by Ca2+.

To elucidate the possible participation of other HMGB1binding proteins in the Ca2+-promoted binding of extracel-lular HMGB1 to the gliosome membrane, we tested theinvolvement of RAGE. A recent report showed that braincontains several splice variants of RAGE, located in thetransmembrane or cytoplasm (Ding and Keller 2005). Wehence evaluated, at first, the amount of membrane-boundRAGE in gliosome and synaptosome preparations. As shownin Fig. 8a, gliosome membranes showed higher levels ofRAGE than synaptosomes. The quantification of immunore-active signals, obtained with different preparations, revealeda 2.7–3.0-fold higher amount of membrane-bound RAGE ingliosomes than in synaptosomes. We could not use a co-immunoprecipitation strategy to measure the effect ofCa2+ on HMGB1/RAGE interaction, as the electrophoreticmobility of full-length transmembrane RAGE in western

Synaptoso

mes

Glioso

mes

GLAST GLAST

(a)

(c)

(b)

GLAST

GLT1

GLT1

IgG heavy chain

IgG heavy chain

IgG light chain

Ip HMGB1

Ip HMGB1

Ip GLASTHMGB1

GLT1

1 2 3 4

21

Fig. 7 Glutamate transporters distribution and HMGB1 binding prop-

erties evaluated by western blotting. (a) GLAST and GLT1 amounts

were measured on 5 lg of hippocampus synaptosome and gliosome

proteins. (b) GLAST and GLT1 levels were evaluated on 5 lg deter-

gent-soluble membrane proteins of hippocampus gliosomes in

immunoprecipitation buffer containing 1.2 mM CaCl2 (lane 1) or 5 mM

EDTA (lane 2). (c) HMGB1 was added to detergent-soluble membrane

extracts and immunoprecipitation of HMGB1 (IP HMGB1) or GLAST

(IP GLAST) was carried out as specified in Materials and methods.

Top and middle panel: Evaluation of GLAST and GLT1 co-immuno-

precipitated (lanes 1 and 3) and not co-immunoprecipitated (lanes 2

and 4) with HMGB1, in the presence of 1.2 mM CaCl2 (lanes 1 and 2)

or 5 mM EDTA (lanes 3 and 4). Bottom panel: evaluation of HMGB1

co-immunoprecipitated (lanes 1 and 3) and not co-immunoprecipitated

(lanes 2 and 4) with GLAST, in the presence of 1.2 mM CaCl2 (lanes 1

and 2) or 5 mM EDTA (lanes 3 and 4). A representative experiment (of

three) is shown.

RAGE

RAGE

RAGE

Synaptoso

mes

Glioso

mes

21

(a)

(b)

(c) 21

Fig. 8 RAGE distribution and HMGB1 binding properties evaluated by

western blotting. (a) RAGE amounts were measured on 5 lg deter-

gent-soluble membrane proteins of hippocampus synaptosomes and

gliosomes. (b) RAGE levels detected using 5 lg detergent-soluble

membrane proteins of hippocampus gliosomes in the presence of

1.2 mM CaCl2 (lane 1) or 5 mM EDTA (lane 2). (c) Evaluation of RAGE

bound to Sepharose-HMGB1 in the presence of 1.2 mM CaCl2 (lane 1)

or 5 mM EDTA (lane 2). A representative experiment (of three) is

shown.

834 M. Pedrazzi et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 99, 827–838� 2006 The Authors

blotting was superimposed on that of antibody heavy chain(not shown). Therefore, HMGB1 was covalently bound toSepharose� beads and its RAGE-binding properties meas-ured in the absence or presence of Ca2+ using detergent-soluble membrane fractions extracted from hippocampalgliosomes. Similar amounts of RAGE were detected in theCa2+-containing (lane 1) and in the Ca2+-free (lane 2) extracts(Fig. 8b). However, RAGE was preferentially boundby HMGB1 in the presence of Ca2+ (Fig. 8c, lane 1) thanin the absence of the ion (Fig. 8c, lane 2). A quantitativecomparison of RAGE immunoreactive material eluted fromSepharose-HMGB1 in both conditions revealed that theamount of RAGE retained by HMGB1 in the presence ofCa2+ was 3.5–4-fold higher than in its absence. These resultssuggest that HMGB1 can interact with both RAGE andGLAST on the gliosome surface but that the preferentialinteraction of HMGB1 with gliosomes in the presence ofCa2+ is driven by RAGE.

Discussion

The present report identifies extracellular HMGB1 as astimulator of glutamate/aspartate release from gliosomes.Specifically, gliosomes, isolated from different brain regionsand labelled with [3H]D-Asp as a stable glutamate analogue,showed an HMGB1-induced concentration-dependent effluxof the amino acid tracer, with a maximal effect detectable inthe presence of nanomolar amounts of external HMGB1.This effect was restricted to glial particles, whereas, in thisrespect, synaptosomes from the same areas of the brain werenot susceptible to HMGB1 at any concentration tested.

Both functional and biochemical evidence supports thenotion that our gliosomal fraction is largely free fromsynaptosomes and the synaptosomal fraction largely freefrom gliosomes. Indeed, the data obtained in confocalmicroscopy and western blotting experiments showing thatsynaptosomes were poorly contaminated by glial-derivedparticles, are mirrored by the release data, which revealed nopotentiation of [3H]D-Asp release by HMGB1 from purifiedsynaptosomes, but only from gliosomes. Consequently,gliosomes do not seem to be noticeably represented in thesynaptosomal fraction. Similarly, gliosomes clearly con-tained only a minority of particles of neuronal origin, ashighlighted in biochemical experiments. Hence, the possiblecontamination by synaptosomes, which do not respond toHMGB1, could not have affected the results obtained.Several studies have exploited the characteristics of thegliosome preparation to study functional aspects of glial cellswith respect to neurones. These studies allowed identificationof specific cell distribution, function and molecular mecha-nisms of a number of transmitter targets, mainly membranetransporters (see for instance Hirst et al. 1998; Daniels andVickroy 1999; Suchak et al. 2003; Raiteri et al. 2005). In avery recent paper we characterised in detail gliosomes

purified from adult rat cerebral cortex and we found themcompetent to actuate glutamate exocytosis when subjected tostimuli able to increase the cytosolic Ca2+ concentration(Stigliani et al. 2006). In keeping with these observations,the present results underline that gliosomes are a vital in vitroglial preparation, able to respond selectively to exogenousstimuli and to reveal subtle differences from their neuronalcounterpart.

The HMGB1-induced [3H]D-Asp release pattern suggeststhat only gliosomes are equipped with functional mediatorsof HMGB1 signalling. To examine the mechanisms promo-ting the release of [3H]D-Asp from gliosomes, we firstanalysed the involvement of external and cytosolic Ca2+ ions.Although the presence of Ca2+ in the external medium isrequired to promote association of HMGB1 with gliosomes,the release of [3H]D-Asp occurs independently of thecytosolic availability of the ion; this may imply activationof carrier-mediated mechanisms. This conclusion is suppor-ted by the ability of DL-TBOA, a potent glutamate uptakeinhibitor able to block all the identified glutamate transportertypes (Shimamoto et al. 1998), to inhibit the HMGB1releasing effect.

Radioactive D-aspartate is commonly used in release anduptake experiments to mimic glutamate in neurones. As[3H]D-Asp cannot be metabolised, one major advantage isthat direct measurement of the radioactive analogue repre-sents that of the authentic molecule. However, the use of thisradioactive analogue has been sometimes viewed withscepticism (Naito and Ueda 1983; Nicholls 1989; Fykseet al. 1992; but see Fleck et al. 1993; Pende et al. 1993;Gundersen et al. 1998). Most important, much less informa-tion is available about gliosomes. In very recent experimentswe found that ionomycin and other releasing stimuli canprovoke exocytotic release of both previously taken up[3H]D-Asp and endogenously synthesised glutamate fromgliosomes; the release of the two species being comparable interms of Ca2+- and transporter-dependence (Stigliani et al.2006; Bonanno et al., unpublished observations). Because ofthe above issues, we decided to perform a set of experimentsmeasuring the effect of HMGB1 on the release of endog-enous glutamate from hippocampal gliosomes. The observa-tion that, under our experimental conditions, HMGB1increased endogenous glutamate release and that this poten-tiation appears to be Ca2+- and DL-TBOA-sensitive, supportsthe results obtained using [3H]D-Asp and suggests that thetracer may reflect the behaviour of the endogenous trans-mitter.

Interestingly, DHK, a selective blocker of GLT1, didnot affect HMGB1 effects. As GLAST and GLT1 are theonly glutamate transporters reported to be expressed inglial cells (Sims and Robinson 1999), our observation ofthe lack of effect of DHK supports the involvement ofGLAST in the increased glutamate release induced byHMGB1. In addition, we demonstrate here that GLAST,

HMGB1-promoted glutamate release in gliosomes 835

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 99, 827–838

which is expressed exclusively by glial phenotypes in thecentral nervous system (Anderson and Swanson 2000),displays HMGB1-binding properties. The finding thatGLAST can be co-immunoprecipitated with HMGB1 sug-gests that the efflux of glutamate from gliosomes bytransporter reversal requires association between HMGB1and GLAST. Moreover, the specificity of this protein–proteininteraction is supported by the absence of the other majorglial transporter of glutamate, GLT1, in the HMGB1 co-immunoprecipitated material. These findings, together withthose of DHK in release studies, strongly suggest that thetransporter involved in the carrier-mediated HMGB1-induced glutamate release from gliosomes may indeed beGLAST.

It is worthy of note that the interaction between GLASTand HMGB1 was not affected by Ca2+, whereas, in contrast,that between RAGE and HMGB1, as well as the binding ofHMGB1 to gliosomal membranes, was promoted by this ion.As the lack of extracellular Ca2+ also inhibited HMGB1-evoked [3H]D-Asp and endogenous glutamate release, thisstrongly suggests that binding of HMGB1 to RAGE plays akey role in the signalling process leading to neurotransmitterrelease.

RAGE is a developmentally regulated protein, and hasbeen proposed as a mediator for HMGB1 signalling in thecentral and peripheral nervous systems (Hori et al. 1995;Rong et al. 2004). Neuronal RAGE has been found prefer-entially located on the cell body surface (Chou et al. 2004)and, in line with this finding, our particles prepared fromneurone endings showed lower levels of RAGE in compar-ison with gliosomes. The ability of HMGB1 simultaneouslyto bind different protein targets on the cell surface in theformation of signalling complexes has been proposedpreviously (Huttunen and Rauvala 2004). In this respect,RAGE could be involved solely as a docking carrier ofextracellular HMGB1 to the cell plasma membrane or as atrue mediator of HMGB1 signalling. In any case, thespecificity of gliosome responses to HMGB1 stimulusultimately depends on the exclusive GLAST expression inglial phenotypes.

Although HMGB1 is widely expressed in the immaturecentral nervous system, the areas which maintain thehighest levels of HMGB1 in adult animals are thehippocampus and cerebellum, where HMGB1 has alsobeen found in the extracellular space (Guazzi et al. 2003).Interestingly, the same areas show the highest levels ofGLAST (Schmitt et al. 1997), suggesting that an interactionbetween this glutamate transporter and the cytokine canactually occur. Recently, it has been shown that GLT1 isfunctionally expressed not only in glial cells but also at thenerve terminal level in adult rodent brain, and that thistransporter activity accounts for the uptake of excitatoryamino acids in both neuronal and non-neuronal phenotypes(Matsui and Jahr 2004). The selective interaction of

HMGB1 with GLAST could therefore be regarded as amechanism that can account for the specific modulation ofglutamate release in gliosomes.

It has been demonstrated that following acute injury to thebrain, such as cerebral ischaemia, stroke or head trauma, glialglutamate transporters contribute to excitotoxic events byreleasing glutamate into the extracellular space through areversed operation of glutamate transport (Rossi et al. 2000;Dunlop et al. 2003). Moreover, it has been recently reportedthat the HMGB1 level was significantly increased in thebrain of Alzheimer disease patients and that co-injection ofHMGB1 delayed the clearance of Ab42 and acceleratedneurodegeneration in Ab42-injected rats (Takata et al. 2004).At present we can only speculate about the role that thealtered activity of GLAST promoted by HMGB1 could playin neurodegenerative and neuro-inflammatory processes.Different glial (Passalacqua et al. 1998) and neuronal (Fageset al. 2000) phenotypes are able to export HMGB1; alter-natively, the extracellular level of the cytokine can beincreased by passive leakage from necrotic cells (Takataet al. 2004). It can be hypothesised that the extracellularrelease of HMGB1 in brain regions of adult mammalsexpressing RAGE and GLAST can promote a local increasein excitotoxicity.

Acknowledgements

This work was supported in part by the Italian Ministero

dell’Istruzione, dell’Universita e della Ricerca (PRIN 2004), FIRB

(Post-Genoma and Neuroscience Projects) and University of Genoa.

The authors are indebted to Mrs Maura Agate for her skilful

secretarial assistance.

References

Agnello D., Wang H., Yang H., Tracey K. J. and Ghezzi P. (2002)HMGB-1, a DNA-binding protein with cytokine activity, inducesbrain TNF and IL-6 production, and mediates anorexia and tasteaversion. Cytokine 18, 231–236.

Anderson C. M. and Swanson R. A. (2000) Astrocyte glutamate trans-port: review of properties, regulation, and physiological functions.Glia 32, 1–14.

Arriza J. L., Fairman W. A., Wadiche J. I., Murdoch G. H., KavanaughM. P. and Amara S. G. (1994) Functional comparisons of threeglutamate transporter subtypes cloned from human motor cortex.J. Neurosci. 14, 5559–5569.

Attwell D., Barbour B. and Szatkowski M. (1993) Nonvesicular releaseof neurotransmitter. Neuron 11, 401–407.

Berridge M. J. (1998) Neuronal calcium signaling. Neuron 21, 13–26.Bustin M. (1999) Regulation of DNA-dependent activities by the

functional motifs of the high-mobility-group chromosomal pro-teins. Mol. Cell. Biol. 19, 5237–5246.

Chou D. K., Zhang J., Smith F. I., McCaffery P. and Jungalwala F. B.(2004) Developmental expression of receptor for advanced gly-cation end products (RAGE), amphoterin and sulfoglucuronyl(HNK-1) carbohydrate in mouse cerebellum and their role inneurite outgrowth and cell migration. J. Neurochem. 90, 1389–1401.

836 M. Pedrazzi et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 99, 827–838� 2006 The Authors

Daniels K. K. and Vickroy T. W. (1999) Reversible activation of glu-tamate transport in rat brain glia by protein kinase C and an oka-daic acid-sensitive phosphoprotein phosphatase. Neurochem. Res.24, 1017–1025.

Ding Q. and Keller J. N. (2005) Evaluation of rage isoforms, ligands,and signaling in the brain. Biochim. Biophys. Acta 1746, 18–27.

Dunlop J., Beal McIlvain H., She Y. and Howland D. S. (2003) Impairedspinal cord glutamate transport capacity and reduced sensitivity toriluzole in a transgenic superoxide dismutase mutant rat model ofamyotrophic lateral sclerosis. J. Neurosci. 23, 1688–1696.

Fages C., Nolo R., Huttunen H. J., Eskelinen E. and Rauvala H. (2000)Regulation of cell migration by amphoterin. J. Cell Sci. 113, 611–620.

Fleck M. W., Henze D. A., Barrionuevo G. and Palmer A. M. (1993)Aspartate and glutamate mediate excitatory synaptic transmissionin area CA1 of the hippocampus. J. Neurosci. 13, 3944–3955.

Fykse E. M., Iversen E. G. and Fonnum F. (1992) Inhibition of1-glutamate uptake into synaptic vesicles. Neurosci. Lett. 135,125–128.

Gray E. G. and Whittaker V. P. (1962) The isolation of nerve endingsfrom brain: an electron-microscopic study of cell fragmentsderived by homogenization and centrifugation. J. Anat. 96, 79–87.

Guazzi S., Strangio A., Franzi A. T. and Bianchi M. E. (2003) HMGB1,an architectural chromatin protein and extracellular signaling fac-tor, has a spatially and temporally restricted expression pattern inmouse brain. Gene Expr. Patterns 3, 29–33.

Gundersen V., Chaudry F. A., Bjaalie J. G., Fonnum F., Ottersen O. P.and Storm-Mathisen J. (1998) Synaptic vesicular localization andexocytosis of 1-aspartate in excitatory nerve terminals: a quanti-tative immunogold analysis in rat hippocampus. J. Neurosci. 18,6059–6070.

Hirst W. D., Price G. W., Rattray M. and Wilkin G. P. (1998) Serotonintransporters in adult rat brain astrocytes revealed by [3H]5-HTuptake into glial plasmalemmal vesicles. Neurochem. Int. 33, 11–22.

Hori O., Brett J., Slattery T., et al. (1995) The receptor for advancedglycation end products (RAGE) is a cellular binding site for am-photerin. Mediation of neurite outgrowth and co-expression ofRAGE and amphoterin in the developing nervous system. J. Biol.Chem. 270, 25 752–25 761.

Huttunen H. J. and Rauvala H. (2004) Amphoterin as an extracellularregulator of cell motility: from discovery to disease. J. Intern. Med.255, 351–366.

Huttunen H. J., Fages C. and Rauvala H. (1999) Receptor for advancedglycation end products (RAGE)-mediated neurite outgrowth andactivation of NF-kappaB require the cytoplasmic domain of thereceptor but different downstream signaling pathways. J. Biol.Chem. 274, 19 919–19 924.

Levi G. and Raiteri M. (1993) Carrier-mediated release of neurotrans-mitters. Trends Neurosci. 16, 415–419.

Matsui K. and Jahr C. E. (2004) Differential control of synaptic andectopic vesicular release of glutamate. J. Neurosci. 24, 8932–8939.

Messmer D., Yang H., Telusma G., Knoll F., Li J., Messmer B., TraceyK. J. and Chiorazzi N. (2004) High mobility group box protein 1:an endogenous signal for dendritic cell maturation and Th1polarization. J. Immunol. 173, 307–313.

Milev P., Chiba A., Haring M., Rauvala H., Schachner M., Ranscht B.,Margolis R. K. and Margolis R. U. (1998) High affinity bindingand overlapping localization of neurocan and phosphacan/protein-tyrosine phosphatase-zeta/beta with tenascin-R., amphoterin, andthe heparin-binding growth-associated molecule. J. Biol. Chem.273, 6998–7005.

Muller S., Ronfani L. and Bianchi M. E. (2004) Regulated expressionand subcellular localization of HMGB1, a chromatin protein with acytokine function. J. Intern. Med. 255, 332–343.

Naito S. and Ueda T. (1983) Adenosine triphosphate-dependent uptakeof glutamate into protein I-associated synaptic vesicles. J. Biol.Chem. 258, 696–699.

Nakamura Y., Iga K., Shibata T., Shudo M. and Kataoka K. (1993)Glial plasmalemmal vesicles: a subcellular fraction from rathippocampal homogenate distinct from synaptosomes. Glia 9, 48–56.

Nakamura Y., Kubo H. and Kataoka K. (1994) Uptake of transmitteramino acids by glial plasmalemmal vesicles from different regionsof rat central nervous system. Neurochem. Res. 19, 1145–1150.

Nicholls D. G. (1989) Release of glutamate, aspartate, and gamma-aminobutyric acid from isolated nerve terminals. J. Neurochem. 52,331–341.

O’Connor K. A., Hansen M. K., Rachal Pugh C., et al. (2003) Furthercharacterization of high mobility group box 1 (HMGB1) as aproinflammatory cytokine: central nervous system effects. Cytokine24, 254–265.

Passalacqua M., Patrone M., Picotti G. B., Del Rio M., Sparatore B.,Melloni E. and Pontremoli S. (1998) Stimulated astrocytes releasehigh-mobility group 1 protein, an inducer of LAN-5 neuroblastomacell differentiation. Neuroscience 82, 1021–1028.

Pende M., Lanza M., Bonanno G. and Raiteri M. (1993) Release ofendogenous glutamic and aspartic acids from cerebrocortex syna-ptosomes and its modulation through activation of a gamma-ami-nobutyric acidB (GABAB) receptor subtype. Brain Res. 604, 325–330.

Raiteri L. and Raiteri M. (2000) Synaptosomes still viable after 25 yearsof superfusion. Neurochem. Res. 25, 1265–1274.

Raiteri M., Angelini F. and Levi G. (1974) A simple apparatus forstudying the release of neurotransmitters from synaptosomes. Eur.J. Pharmacol. 25, 411–414.

Raiteri M., Sala R., Fassio A., Rossetto O. and Bonanno G. (2000)Entrapping of impermeant probes of different size into nonper-meabilized synaptosomes as a method to study presynapticmechanisms. J. Neurochem. 74, 423–431.

Raiteri L., Stigliani S., Siri A., Passalacqua M., Melloni E., Raiteri M.and Bonanno G. (2005) Glycine taken up through GLYT1 andGLYT2 heterotransporters into glutamatergic axon terminals ofmouse spinal cord elicits release of glutamate by homotransporterreversal and through anion channels. Biochem. Pharmacol. 69,158–168.

Rong L. L., Trojaborg W., Qu W., Kostov K., Yan S. D., Gooch C.,Szabolcs M., Hays A. P. and Schmidt A. M. (2004) Antagonism ofRAGE suppresses peripheral nerve regeneration. FASEB J. 18,1812–1817.

Rossi D. J., Oshima T. and Attwell D. (2000) Glutamate release insevere brain ischemia is mainly by reversed uptake. Nature 403,316–321.

Sajithlal G., Huttunen H., Rauvala H. and Munch G. (2002) Receptor foradvanced glycation end products plays a more important role incellular survival than in neurite outgrowth during retinoic acid-induced differentiation of neuroblastoma cells. J. Biol. Chem. 277,6888–6897.

Schmitt A., Asan E., Puschel B. and Kugler P. (1997) Cellular andregional distribution of the glutamate transporter GLAST in theCNS of rats: nonradioactive in situ hybridization and comparativeimmunocytochemistry. J. Neurosci. 17, 1–10.

Semino C., Angelini G., Poggi A. and Rubartelli A. (2005) NK/iDCinteraction results in IL-18 secretion by DCs at the synaptic cleftfollowed by NK cell activation and release of the DC maturationfactor HMGB1. Blood 106, 609–616.

HMGB1-promoted glutamate release in gliosomes 837

� 2006 The AuthorsJournal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 99, 827–838

Shimamoto K., Lebrun B., Yasuda-Kamatani Y., Sakaitani M., ShigeriY., Yumoto N. and Nakajima T. (1998) DL-threo-beta-benzyl-oxyaspartate, a potent blocker of excitatory amino acid transport-ers. Mol. Pharmacol. 53, 195–201.

Sims K. D. and Robinson M. B. (1999) Expression patterns and regu-lation of glutamate transporters in the developing and adult nervoussystem. Crit. Rev. Neurobiol. 13, 169–197.

Sparatore B., Passalacqua M., Patrone M., Melloni E. and Pontremoli S.(1996) Extracellular high-mobility group 1 protein is essential formurine erythroleukaemia cell differentiation. Biochem. J. 320,253–256.

Sparatore B., Patrone M., Passalacqua M., Pedrazzi M., Ledda S.,Pontremoli S. and Melloni E. (2005) Activation of A431 humancarcinoma cell motility by extracellular high mobility group box 1protein and epidermal growth factor stimuli. Biochem. J. 389, 215–221.

Stigliani S., Zappettini S., Raiteri L., Passalacqua M., Melloni E., VenturiC., Tacchetti C., Diaspro A., Usai C. and Bonanno G. (2006) Glia

re-sealed particles freshly prepared from adult rat brain are com-petent for exocytotic release of glutamate. J. Neurochem. 96, 656–668.

Suchak S. K., Baloyianni N. V., Perkinton M. S., Williams R. J., Mel-drum B. S. and Rattray M. (2003) The ‘glial’ glutamate transporter,EAAT2 (Glt-1) accounts for high affinity glutamate uptake intoadult rodent nerve endings. J. Neurochem. 84, 522–532.

Taguchi A., Blood D. C., del Toro G., et al. (2000) Blockade of RAGE-amphoterin signaling suppresses tumour growth and metastases.Nature 405, 354–360.

Takata K., Kitamura Y., Tsuchiya D., Kawasaki T., Taniguchi T. andShimohama S. (2004) High mobility group box protein-1 inhibitsmicroglial Abeta clearance and enhances Abeta neurotoxicity.J. Neurosci. Res. 78, 880–891.

Wang H., Vishnubhakat J. M., Bloom O., Zhang M., Ombrellino M.,Sama A. and Tracey K. J. (1999) Proinflammatory cytokines (tu-mor necrosis factor and interleukin 1) stimulate release of highmobility group protein-1 by pituicytes. Surgery 126, 389–392.

838 M. Pedrazzi et al.

Journal Compilation � 2006 International Society for Neurochemistry, J. Neurochem. (2006) 99, 827–838� 2006 The Authors