Different roles of adenosine A1, A2A and A3 receptors in controlling kainate-induced toxicity in cortical cultured neurons

www.elsevier.com/locate/neuint

Neurochemistry International 47 (2005) 317–325

Different roles of adenosine A1, A2A and A3 receptors in controlling

kainate-induced toxicity in cortical cultured neurons

Nelson Rebola, Ricardo J. Rodrigues, Catarina R. Oliveira, Rodrigo A. Cunha *

Center for Neurosciences of Coimbra, Institute of Biochemistry, Faculty of Medicine, University of Coimbra, 3004-504 Coimbra, Portugal

Received 29 March 2005; received in revised form 29 April 2005; accepted 13 May 2005

Abstract

Adenosine is a neuromodulator that can control brain damage through activation of A1, A2A and A3 receptors, which are located in both

neurons and other brain cells. We took advantage of cultured neurons to investigate the role of neuronal adenosine receptors in the control of

neurotoxicity caused by kainate and cyclothiazide. Both A1, A2A and A3 receptors were immunocytochemically identified in cortical neurons.

Activation of A1 receptors with 100 nM CPA did not modify the extent of neuronal death whereas the A1 receptor antagonist, DPCPX

(50 nM), attenuated neurotoxicity by 28 � 5%, and effect similar to that resulting from the removal of endogenous adenosine with 2 U/ml of

adenosine deaminase (27 � 3% attenuation of neurotoxicity). In the presence of adenosine deaminase, DPCPX had no further effect and CPA

now exacerbated neurotoxicity by 42 � 4%. Activation of A2A receptor with 30 nM CGS21680 attenuated neurotoxicity by 40 � 8%, an

effect prevented by the A2A receptor antagonists, SCH58261 (50 nM) or ZM241385 (50 nM), which by themselves were devoid of effect.

Finally, neither A3 receptor activation with Cl-IB-MECA (100–500 nM) nor blockade with MRS1191 (5 mM) modified neurotoxicity. These

results show that A1 receptor activation enhances and A2A receptor activation attenuates neurotoxicity in cultured cortical neurons, indicating

that these two neuronal adenosine receptors directly control neurodegeneration. Interestingly, the control by adenosine of neurotoxicity in

cultured neurons is similar to that observed in vivo in newborn animals and is the opposite of what is observed in adult brain preparations

where A1 receptor activation and A2A receptor blockade are neuroprotective.

# 2005 Elsevier Ltd. All rights reserved.

Keywords: Adenosine; Neuroprotection; A1 receptor; A2A receptor; A3 receptor

1. Introduction

Adenosine is a ubiquitous neuromodulator in the central

nervous system acting via four G-protein coupled (A1, A2A,

A2B and A3) receptors (Fredholm et al., 2005). Adenosine is

considered a neuroprotective substance, which is able to

prevent or decrease neuronal damage in the adult brain in

different noxious brain conditions such as hypoxia/ischemia,

excitotoxicity, chemotoxicity or trauma (de Mendonca et al.,

2000). In fact, adenosine is released upon conditions of

metabolic stress and many of the known effects of adenosine

are indeed compatible with neuroprotective properties, since

adenosine is able to decrease the release of excitatory amino

* Corresponding author. Tel.: +351 239 820190; fax: +351 239 822776.

E-mail address: [email protected] (R.A. Cunha).

URL: http://cnc.cj.uc.pt/lab_lef/

0197-0186/$ – see front matter # 2005 Elsevier Ltd. All rights reserved.

doi:10.1016/j.neuint.2005.05.009

acids, hyperpolarize neurons, restrain the activation of

N-methyl-D-aspartate (NMDA) receptors, limit calcium

influx, inhibit free radical formation, and exert modulatory

effects in astrocytes and microglia (Fredholm, 1997). Given

the effects mediated by the activation of adenosine A1

receptors, namely to inhibit neuronal Ca2+ transients, to

reduce excitatory amino acid release and to hyperpolarize

neurons, most of the neuroprotective effects of adenosine in

the adult brain are thought to be mediated by neuronal A1

receptors (Fredholm et al., 2005). Accordingly, the acute

administration of A1 receptor agonists affords brain

neuroprotection and, conversely, A1 receptor antagonists

exacerbate brain damage in adult animals (reviewed by de

Mendonca et al., 2000).

Recently, a series of observations have highlighted the

role of another adenosine receptor, the A2A receptor, in

controlling brain neuronal damage (reviewed by Cunha,

N. Rebola et al. / Neurochemistry International 47 (2005) 317–325318

2005). Despite its low abundance in brain regions other than

the striatum (e.g. Alfaro et al., 2004; Lopes et al., 2004),

blocking A2A receptors with antagonists or genetically

deleting the A2A receptor gene results in a decrease of the

extent of neuronal damage and neurological deficits in

different noxious brain conditions in adult animals

(reviewed by Cunha, 2005). However, the brain neuropro-

tection afforded by blockade of A2A receptors in adult

animals is mostly phenomenological since the mechanisms

operated by A2A receptors to confer such a robust

neuroprotection are not known. In particular, since A2A

receptors are located in neurons but also in astrocytes,

microglia cells and brain capillaries (reviewed by Fredholm

et al., 2005), the neuroprotective effect resulting from A2A

receptor blockade in adult animals might involve either

neuronal or non-neuronal A2A receptors.

A similar situation occurs with respect to the possible role

of adenosine A3 receptors in the control of brain neuronal

damage. In fact, some reports support a possible role for this

receptor in controlling brain damage (von Lubitz et al.,

1994; Fedorova et al., 2003; Pugliese et al., 2003). However,

given the low abundance of this receptor in neurons (Diaz-

Hernandez et al., 2002; Lopes et al., 2003a) and its lack of

significant effects in modulating synaptic transmission or

neuronal excitability (Brand et al., 2001; Lopes et al.,

2003b), its role in controlling neuronal function is still

unclear. Since A3 receptors are also located in astrocytes

(e.g. Brambilla et al., 2003) and in microglia cells

(Hammarberg et al., 2003), it is possible that the impact

of these receptors on neuronal damage might indirectly

result from the control of the reactivity of glia cells.

As a first step to understand if the neuroprotective

properties of adenosine receptors might be due to a direct

control by neuronal adenosine receptors of neuronal

viability, we took advantage of an enriched preparation of

neurons to test the direct neuroprotective properties of

neuronal A1, A2A and A3 receptors. Thus, we tested the

effect of A1, A2A and A3 receptor agonists and antagonists

on a particular excitotoxicity model in rat cortical cultured

neurons, induced by kainate + cyclothiazide (e.g. Ambrosio

et al., 2000).

2. Experimental procedures

2.1. Chemicals

1,3-Dipropyl-8-cyclopentyladenosine (DPCPX) and

N6-cyclopentyladenosine (CPA) were from RBI (Sigma–

Aldrich Quımica, Portugal), 3-ethyl-5-benzyl-2-methyl-4-

phenylethynyl-6-phenyl-1,4-dihydropyridine-3,5-dicarboxy-

late (MRS 1191), 2-chloro-N6-(3-iodobenzyl)adenosine-50-N-methyluronamide (Cl-IB-MECA), 2-[p-(2-carboxyethyl)-

phenylethylamino]-50-N-ethylcarboxamidoadenosine (CGS

21680), 4-(2-[7-amino)]-2-(2-furyl{1,2,4}-triazolo{2,3-

a{1,3,5}triazin-5-yl-aminoethyl)phenol} (ZM 241385),

kainate and cyclothiazide (CTZ) were from Tocris Cookson

(Bristol, UK), adenosine deaminase (type VI, 2000 U/ml)

was from Sigma (Reagente 5, Portugal) and 5-amino-7-(2-

phenylethyl)-2-(2-furyl)-pyrazolo[4,3-e]-1,2,4-triazolo[1,5-

c]pyrimidine (SCH 58261) was a generous gift from

S. Weiss (Vernalis, UK). Rabbit purified IgG anti-adenosine

A1 receptor antibody (1.8 mg/ml) was from Affinity

Bioreagents (Golden, CO, USA), goat anti-adenosine A2A

receptor antibody (200 mg/ml) was from Santa Cruz

Biotechnology-Europe (Freelab, Lisbon, Portugal), and rabbit

anti-adenosine A3 receptor antibody (1 mg/ml) was from

Chemicon (Temecula, CA, USA). Cell culture solutions were

purchased from GIBCO BRL (Life Technologies, Scotland,

UK).

CGS 21680, Cl-IB-MECA, CPA, MRS 1191, SCH

58261, ZM 241385, kainate and CTZ were made up to a

5 mM stock solution in dimethylsulfoxide and DPCPX was

made up into a 5 mM stock in 99% dimethylsulfoxide and

1% NaOH 1 M. These stock solutions were aliquoted and

stored at �20 8C and aqueous dilution of these stock

solutions was made daily.

2.2. Preparation of rat brain cortical neurons

Primary cultures of cortical neurons were obtained from

15 to 16 days embryos of Wistar rats as previously described

(Agostinho and Oliveira, 2003). Briefly, the neocortices of

the embryos were dissected and placed in Ca2+ and Mg2+-

free Krebs buffer (mM): NaCl, 120; KCl, 4.8; KH2PO4, 1.2;

glucose, 13; HEPES, 10 (pH 7.4) and 0.3% bovine serum

albumin. After removal of the meninges, the tissue was

washed and incubated in Krebs buffer containing 0.02%

trypsin and 0.004% DNAse I, for 10 min at 37 8C. The

digestion was stopped with Krebs buffer containing 0.05%

trypsin inhibitor (type II-S) and 0.004% DNAse I and the

tissue was centrifuged at 140 � g for 5 min. After washing

the pellet once with Krebs buffer, the cells were dissociated

mechanically through a large-bore 5 ml glass pipette.

Cortical neurons were cultured in Neurobasal medium

supplemented with 2 mM L-glutamine, 2% B27 supplement,

penicillin (100 U/ml) and streptomycin (100 mg/ml). The

cells were plated in poly-L-lysine (0.1 mg/ml) coated cover-

slips at a density of 1 � 105 cells per cover-slip and the

cultures were maintained at 37 8C in a humidified atmo-

sphere of 5% CO2/95% air. In these conditions, we obtained

a near 90% pure neuronal preparation, as evaluated by the

ratio of elements stained with antibodies against a

microtubule-associated protein (MAP-2, a neuronal marker)

and glial fibrillary acidic protein (GFAP, an astrocytic

marker).

2.3. Immunocytochemistry

Immunocytochemistry in the cover-slip-mounted neu-

rons was carried out as previously described (Rebola et al.,

2005). Briefly, the cultures were washed twice with 1 ml

N. Rebola et al. / Neurochemistry International 47 (2005) 317–325 319

PBS (140 mM NaCl, 3 mM KCl, 20 mM Na2HPO4, 1.5 mM

KH2PO4), kept at 37 8C during 10 min, and fixed with 4%

paraformaldehyde with 4% sucrose for 30 min at 37 8C.

Cover-slip-mounted cells were then permeabilized with

0.2% Triton X-100 for 2 min at room temperature and non-

specific binding subsequently blocked with 3% bovine

serum albumin for 30 min at room temperature. Cells were

then incubated for 1 h at room temperature with primary

antibodies, namely goat anti-adenosine A2A receptor anti-

body (1:400 dilution), rabbit anti-A3 receptor antibody

(1:400 dilution) or rabbit anti-A1 receptor antibody (1:300

dilution). After washing three times with 200 ml PBS for

5 min, the cells were incubated with the complementary

secondary antibodies, an AlexaFluor-488 (green) labelled

donkey anti-goat IgG antibody (Amersham, 1 mg/ml, 1:200

dilution) or an AlexaFluor-488 labelled goat anti-rabbit IgG

antibody (Amersham, 1 mg/ml, 1:200 dilution), during 1 h

at room temperature. After three washing periods of 5 min

with 200 ml PBS, the cells were mounted using a Prolong

Antifade kit (Amersham) and, after drying, were visualized

using a BioRad 600 confocal microscope or a Zeiss Axiovert

200 fluorescence microscope equipped with a cooled CCD

camera and analysed with MetaFluor 4.0 software. The

selectivity of these antibodies has already been validated in

previous studies (Lopes et al., 2003a, 2004; Rebola et al.,

2003, 2005). In particular, we have previously confirmed

that the antibodies used recognize a single band in Western

blot analysis of brain membranes, which disappears in brain

membranes derived from knockouts of the receptor for

which the antibodies are selective (Lopes et al., 2003a, 2004;

Rebola et al., 2003, 2005). Furthermore, we have confirmed

that the immunoreactivity for these antibodies disappears on

pre-incubation with the corresponding immunizing pep-

tides.

2.4. Induction of neuronal death

To induce neuronal death, we used a toxicity model based

on the administration of kainate and cyclothiazide that has

already been described and validated (e.g. Ambrosio et al.,

2000). We selected this particular model to induce

excitotoxicity because it triggers a mild pattern of neuronal

death, which allows studying both protection and exacer-

bation of neurotoxicity by neuromodulators. Thus, our goal

was to compare the influence of different adenosine

receptors on neurotoxicity rather that to characterise the

molecular mechanisms of neurotoxicity caused by admin-

istration of kainate and cyclothiazide, which has previously

been found to activate mainly AMPA receptors to trigger

neurodegeneration of cultured neurons (see Ambrosio et al.,

2000). Briefly, after six days in culture, cortical neurons

were exposed to 100 mM kainate and 30 mM cyclothiazide

(which avoids AMPA receptor desensitization) during 24 h.

These drugs were simultaneously added to the cultures by

removing 500 ml of the medium (out of 1 ml) and adding

500 ml of fresh medium containing kainate and cyclothia-

zide, followed by a gentle homogenization of the medium to

preserve the integrity of neurons. Adenosine receptor

agonists and antagonists were added to the cultures in a

similar manner, 30 min before the excitotoxic stimulus and

were present during the subsequent 24 h incubation period.

2.5. Assessment of neuronal death

The analysis of the extent of neuronal death was assessed

using the fluorescent probes SYTO-13 and propidium iodide

(PI). SYTO-13 labels RNA and DNA in living cells with a

green emission. Propidium iodide is excluded from viable

cells and labels cells with a disrupted plasma membrane with

a red emission. Cells were loaded for 3 min with a solution

of Krebs buffer (132 mM NaCl, 4 mM KCl, 1.4 mM MgCl2,

1 mM CaCl2, 6 mM glucose, 10 mM HEPES-Na, pH 7.4)

containing 4 mM SYTO-13 and 4 mg/ml PI. The cells (viable

and non-viable) were counted using an inverted Nikon

Diaphot fluorescent microscope and neurotoxicity was

expressed as percentage of non-viable cells compared to

the total number of cells after counting the number of

elements in at least three different fields per cover-slip with a

minimum number of 100 elements per cover-slip.

A second methodology was used to confirm the loss of

viability of the cultured cortical neurons based on the extent

of reduction of the Alamar Blue dye (Biosource), which is

proportional to the number of viable cells containing

competent mitochondria. Twenty hours after the beginning

of the incubation with the tested drugs, Alamar Blue

(100 ml) was added to each well containing cultured cortical

neurons. After 4 h, 100 ml aliquots of the media was

collected from each well and fluorescence was measured in a

microplate reader (at 600 nm upon excitation at 570 nm) and

the percentage reduction of viable cells was considered

equivalent to the percentage decrease of the reduction of

Alamar Blue.

2.6. Statistics

Data are mean � SEM values of n experiments (i.e.

cultures derived from different groups of animals).

Significance was considered at P < 0.05 using the Student’s

t-test.

3. Results

3.1. Adenosine A1, A2A and A3 receptors are present in

cultured cortical neurons

We first confirmed if adenosine receptors were present in

cultured cortical neurons of the rat. To address this issue, we

probed cortical cultures with antibodies against the

adenosine A1, A2A and A3 receptors. As illustrated in

Fig. 1, we were able to detect by immunocytochemistry the

presence of these three adenosine receptors. As illustrated in

N. Rebola et al. / Neurochemistry International 47 (2005) 317–325320

Fig. 1. Localization of adenosine A1, A2A and A3 receptor immunoreactivity in rat cortical neurons in primary culture. The immunocytochemistry to reveal

adenosine receptors was carried out using a rabbit anti-A1 receptor antibody (1:300, left panel), a goat anti-A2A receptor antibody (1:400 dilution, middle panel)

or a rabbit anti-A3 receptor antibody (1:400, right panel) and an AlexaFluor-488 (green) labelled goat anti-rabbit IgG antibody (1:200 dilution) or an

AlexaFluor-488 labelled donkey anti-goat IgG antibody (1:200 dilution). These images are representative of three different experiments carried out using three

independent cultures of cortical neurons.

Fig. 1, A1 receptor immunoreactivity was mainly observed

in the cell body region of neurons. A1 receptor

immunoreactivity was also observed with less intensity

in processes and with even less intensity in nerve

terminals, as evaluated by its minor co-localization with

synaptophysin immunoreactivity (a marker of synaptic

vesicles) (data not shown). As illustrated in Fig. 1, A2A

receptor immunoreactivity displayed a punctuated pattern

and was mostly located in nerve terminals, as evaluated by

its co-localization with synaptophysin immunoreactivity

(data not shown), as previously reported to occur in

cultured hippocampal neurons (Rebola et al., 2005). A2A

receptor immunoreactivity was also located with less

intensity in the cell body region, as observed by others

(Lee et al., 2004). Finally, as illustrated in Fig. 1, A3

receptor immunoreactivity was mostly disseminated

throughout the cell body region of cortical neurons and

confocal microscopy analysis revealed a mostly intracel-

lular localization of this receptor.

3.2. Extent of neuronal death induced by kainate and

cyclothiazide

The incubation of primary cultures of cortical neurons

with 100 mM kainate and 30 mM cyclothiazide during 24 h

led to a 42.7 � 4.5% (n = 13) neuronal death, as assessed

using propidium iodide and SYTO-13. These injured

neurons became round shape elements stained red (meaning

that they have lost their structural and membrane integrity

allowing PI to stain the nucleus) in contrast to viable

neurons, which display a typical ramified morphology with a

larger nucleus diffusely stained with SYTO 13, i.e. green.

Using a second independent methodology to evaluate the

loss of cellular viability based on the reduction of Alamar

Blue, we confirmed that exposure of cortical neurons to

kainate + cyclothiazide caused a 40.9 � 3.7% (n = 10)

decrease of viable cells. This extent of toxicity allows

investigating if the activation or blockade of A1, A2A and A3

receptors might either attenuate or exacerbate neurotoxicity.

For that purpose, we selected single concentrations of

selective agonists and antagonists for each of these

adenosine receptors, since our goal was to compare the

role of the different adenosine receptors on neurotoxicity

rather than to define potencies or efficiencies of adenosine

receptor ligands.

3.3. Adenosine A1 receptor blockade is neuroprotective

To investigate the role of A1 receptors, we tested the

effect of the selective A1 receptor agonist, CPA at a

concentration of 100 nM, which is effective to activate A1

receptors in cultured neurons (e.g. Scholz and Miller, 1991)

and is the maximal concentration of CPA that is devoid of

effects through activation of A2A receptors (see Cunha and

Ribeiro, 2000). We found that CPA (100 nM) had no effects

on the neurotoxicity caused by kainate + cyclothiazide in

cultured cortical neurons (Fig. 2). In contrast, the selective

A1 receptor antagonist, DPCPX (50 nM), decreased

neuronal cell death induced by kainate + cyclothiazide to

30.7 � 4.9% (n = 4, P < 0.05 compared to the effect of

kainate + cyclothiazide in the absence of DPCPX; Fig. 2).

This indicates that the tonic activation of A1 receptors by

endogenous adenosine exacerbates neuronal damage trig-

gered by kainate + cyclothiazide. It also suggests that the

levels of endogenous extracellular adenosine might already

saturate A1 receptors thus blunting the effects of exogen-

ously added A1 receptor agonists, as previously observed by

others (Lobner, 2002).

N. Rebola et al. / Neurochemistry International 47 (2005) 317–325 321

Fig. 2. Lack of effect of a selective A1 receptor agonist and neuroprotection

by a selective A1 receptor antagonist in cultured cortical neurons exposed to

kainate and cyclothiazide. Cortical neurons were incubated for 24 h in the

absence (open bars) or in the presence of 100 mM kainate and 30 mM

cyclothiazide (filled bars), which decreased the number of viable by near

40% (compare the first two bars from the left), as evaluated by the number of

cells labelled with propidium iodide. When the same experiment was

performed in the presence of the selective A1 receptor agonist, CPA

(100 nM), there was no modification of the number of viable cells either

in the absence (open bar) or in the presence of kainate and cyclothiazide

(filled bar). Finally, in the presence of the selective A1 receptor antagonist,

DPCPX (50 nM), there was no modification of the number of viable cells in

control conditions (open bar), but the neurotoxicity triggered by kainate and

cyclothiazide was now decreased to 30.7 � 4.9% (filled bar). Each bar is the

mean � SEM of four experiments. *P < 0.05 between the indicated bars or

groups of experiments.

Fig. 3. Neuroprotection by an A2A receptor agonist and lack of effect of A2A

receptor antagonists in cultured cortical neurons exposed to kainate and

cyclothiazide. Cortical neurons were incubated for 24 h in the absence

(open bars) or in the presence of 100 mM kainate and 30 mM cyclothiazide

(filled bars), which decreased the number of viable by near 40% (compare

the first two bars from the left), as evaluated by the number of cells labelled

with propidium iodide. When the same experiment was performed in the

presence of the A2A receptor agonist, CGS 21680 (CGS, 30 nM), there was

no modification of the number of viable cells in the absence of kainate and

cyclothiazide (open bar) but the neurotoxicity triggered by kainate and

cyclothiazide was decreased to 25.8 � 4.7% (filled bar). In contrast, in the

presence of the selective A2A receptor antagonists, ZM 241385 (ZM,

50 nM) or SCH 58261 (SCH, 50 nM), there was no modification of the

number of viable cells neither in the absence (open bars) nor in the presence

of kainate and cyclothiazide (filled bars) compared to control. However, in

the presence of either ZM 241385 or of SCH 58261, the neuroprotection

afforded by CGS 21680 was no longer observed. Each bar is the mean

� SEM of 3–5 experiments. *P < 0.05 between the indicated bars or groups

of experiments.

3.4. Adenosine A2A receptor activation is

neuroprotective

To investigate the role of A2A receptors, we tested the

effect of the best characterised commercially available

agonist of this receptor, CGS 21680. However, it has

previously been shown that CGS 21680 can also bind and

activate A1 receptors (Lupica et al., 1990; Lopes et al.,

2004). But since we had previously found that a selective

A1 receptor agonist (CPA) was devoid of effects in the

control of neuronal viability, we reasoned that any effect of

CGS 21680 in our model should be attributed to A2A

receptor-mediated actions. CGS 21680 was tested at a

concentration of 30 nM, which we (Agostinho et al., 2000)

and others (Ferreira and Paes de Carvalho, 2001) have

found to activate A2A receptors in cultured neurons and

which produces a maximal A2A receptor activation in

limbic cortical preparations (e.g. Cunha et al., 1997). As

illustrated in Fig. 3, CGS 21680 (30 nM) reduced the

neuronal death induced by kainate and cyclothiazide to

25.8 � 4.7% (n = 6, P < 0.05 compared to the effect of

kainate + cyclothiazide in the absence of CGS 21680).

This neuroprotective effect of CGS 21680 was prevented

by the A2A receptor antagonists, ZM 241385 (50 nM,

n = 4) and SCH 58261 (50 nM, n = 3), while none of these

A2A receptor antagonists had any effect on neuronal

viability per se (Fig. 3). This demonstrates that the

neuroprotective effect of CGS 21680 indeed results from

its ability to activate A2A receptors, which do not seem to

be tonically activated by endogenous adenosine.

3.5. Adenosine A3 receptors do not affect neuronal

death

To probe the role of A3 receptors on neuronal viability,

we tested the effect of a purported A3 receptor agonist, Cl-

IB-MECA, and of a selective A3 receptor antagonist, MRS

1191. It has previously been shown that sub-micromolar

concentrations of Cl-IB-MECA can bind and activate A3

receptors in brain preparations (Brand et al., 2001; Costenla

et al., 2001; Diaz-Hernandez et al., 2002), although they can

also bind and activate A1 receptors (Costenla et al., 2001;

Rivkees et al., 2001). However, since a selective A1 receptor

agonist (CPA) was devoid of effects on neuronal viability in

our model, we reasoned that any effect of Cl-IB-MECA

should be attributed to A3 receptor-mediated actions. But, as

shown in Fig. 4, neither the A3 receptor agonist, Cl-IB-

MECA (100 nM, n = 3 or 500 nM, n = 2), nor the A3

receptor antagonist, MRS 1191 (5 mM, n = 3), had any

significant effect on neuronal survival or death triggered by

kainate + cyclothiazide.

N. Rebola et al. / Neurochemistry International 47 (2005) 317–325322

Fig. 4. Lack of effect of a either an A3 receptor agonist or an A3 receptor

antagonist in cultured cortical neurons exposed to kainate and cyclothiazide.

Cortical neurons were incubated for 24 h in the absence (open bars) or in the

presence of 100 mM kainate and 30 mM cyclothiazide (filled bars), which

decreased the number of viable by near 40% (compare the first two bars

from the left), as evaluated by the number of cells labelled with propidium

iodide. When the same experiment was performed in the presence of the A3

receptor agonist, Cl-IB-MECA (100 nM), there was no modification of the

number of viable cells either in the absence (open bar) or in the presence of

kainate and cyclothiazide (filled bar). Likewise, in the presence of the A3

receptor antagonist, MRS 1191 (5 mM), there was no modification of the

number of viable cells either in the absence (open bar) or in the presence of

kainate and cyclothiazide (filled bar). Each bar is the mean � SEM of three

experiments. *P < 0.05 between the indicated bars.

Fig. 5. Removal of endogenous extracellular adenosine with adenosine

deaminase is neuroprotective in cultured cortical neurons exposed to kainate

and cyclothiazide and allows revealing that the activation of A1 receptors

exacerbates neurotoxicity. Cortical neurons were incubated for 24 h in the

absence (open bars) or in the presence of 100 mM kainate and 30 mM

cyclothiazide (filled bars), which decreased the number of viable by near

40% (compare the first two bars from the left), as evaluated by the number of

cells labelled with propidium iodide. When the same experiment was

performed in the presence of adenosine deaminase (2 U/ml), there was

no modification of the number of viable cells in the absence of kainate and

cyclothiazide (open bar) but the neurotoxicity triggered by kainate and

cyclothiazide was decreased to 33.5 � 2.9% (filled bar). This neuroprotec-

tion afforded by adenosine deaminase might result from removal of an A1

receptor tonus since the simultaneous presence of adenosine deaminase

(2 U/ml) and of the A1 receptor antagonist, DPCPX (50 nM) caused a

neuroprotection similar to that observed with adenosine deaminase only.

Finally, in the presence of adenosine deaminase (2 U/ml), the A1 receptor

agonist, CPA (100 nM) now exacerbated neuronal death caused by kainate

and cyclothiazide. Each bar is the mean � SEM of 3–4 experiments.*P < 0.05 between the indicated bars or group of experiments.

3.6. Endogenous adenosine acting through A1 receptors

exacerbates neuronal death

The data obtained so far indicate that the only adenosine

receptor tonically activated by endogenous adenosine under

our experimental conditions is the A1 receptor. In fact,

neuronal death triggered by kainate + cyclothiazide in

cultured cortical neurons was only modified by the A1

receptor antagonist, DPCPX, but not by A2A receptor

antagonists, ZM 241385 or SCH 58261, or by the A3

receptor antagonist, MRS 1191. To confirm this conclusion,

we tested the effect of removing endogenous extracellular

adenosine using adenosine deaminase, which converts

adenosine into its centrally inactive metabolite – inosine.

As illustrated in Fig. 5, adenosine deaminase (2 U/ml)

reduced neuronal death to 33.5 � 2.9% (n = 4, P < 0.05

compared to the effect of kainate + cyclothiazide in the

absence of adenosine deaminase). Thus, the similar

qualitative and quantitative effects of adenosine deaminase

and of DPCPX further support that only A1 receptors are

tonically playing a role in controlling neuronal viability. If

this is correct, one would expect that the neuroprotective

effects of DPCPX and of adenosine deaminase were not

additive. As illustrated in Fig. 5, there was no significant

difference between the neuroprotective effects afforded by

adenosine deaminase, by DPCPX or by their simultaneous

presence.

When exploring the role of A1 receptor agonists in the

control of neuronal viability, the lack of effect of CPA was

interpreted as a ‘saturation’ by endogenous adenosine of the

A1 receptor-mediated control of neuronal survival. This can

now be directly tested by exploring the effect of CPA in the

presence of adenosine deaminase. As illustrated in Fig. 5, it

was observed that CPA increased neuronal damage once the

extracellular levels of adenosine were reduced by adenosine

deaminase. In fact, whereas neuronal death triggered by

kainate and cyclothiazide was 30.7 � 4.9% in the presence

of adenosine deaminase (2 U/ml), it increased to

42.1 � 3.8% (n = 4; P < 0.05) once CPA (100 nM) was

also present.

4. Discussion

The present study constitutes the first attempt to directly

compare the relative roles of neuronal adenosine receptors in

the control of neuronal death using the same excitotoxic

stimulus (kainate + cyclothiazide) in the same model, i.e.

cultured cortical neurons, which were over 90% pure and

were endowed with adenosine A1, A2A and A3 receptors. We

tested single concentrations of agonist and antagonists of

these adenosine receptors with care to choose concentrations

N. Rebola et al. / Neurochemistry International 47 (2005) 317–325 323

that are selective but supra-maximal for the activation or

antagonism of each adenosine receptor. In fact, since our

goal was to compare the role of each adenosine receptor

subtype rather than to define relative potencies or efficacies

of different adenosine receptor ligands, we did not feel the

need to carry out concentration-responses curves for the

chosen adenosine receptor ligands, an option that was

justified by the unambiguously distinct effects resulting

from the manipulation of each adenosine receptor subtype.

The use of selective concentrations of A1 and A2A

receptor agonist and antagonist allowed concluding that the

activation of neuronal A1 receptors exacerbates neurotoxi-

city, whereas the activation of neuronal A2A receptors

affords neuroprotection in cortical cultured neurons. In fact,

the blockade of A1 receptors with DPCPX was neuropro-

tective, indicating that the tonic activation of A1 receptors is

contributing for the exacerbation of kainate + cyclothia-

zide-induced neurotoxicity in cultured cortical neurons.

Our observations are in agreement with studies by others

showing that A1 receptor activation exacerbates neurotoxi-

city in different cultured neurons (Lobner and Choi, 1994;

Barth et al., 1997; Turner et al., 2004; but see Logan and

Sweeney, 1997). Accordingly, we now observed that the

activation of A1 receptors with its selective agonist CPA

exacerbated neurotoxicity provided that endogenous ade-

nosine was removed using adenosine deaminase. This

confirms that the activation of neuronal A1 receptors is

indeed detrimental for cultured neurons. In contrast, we

observed that the activation of neuronal A2A receptors

had an effect opposite to that of A1 receptors, i.e. activation

of A2A receptors afforded neuroprotection against the

kainate + cyclothiazide-induced neurotoxicity in the cul-

tured cortical neurons. This is agreement with the

previously reported neuroprotection resulting from the

activation of A2A receptors in cultured retina neurons (e.g.

Ferreira and Paes de Carvalho, 2001). In marked contrast

with the participation of neuronal A1 and A2A receptor in

the control of neurotoxicity, neuronal A3 receptors failed

to modify the neurotoxicity induced by kainate + cyclothia-

zide suggesting that any eventual control of brain

neurotoxicity by A3 receptors might preferentially involve

A3 receptors located is astrocytes (e.g. Brambilla et al.,

2003) or in microglia (Hammarberg et al., 2003) rather

than neuronal A3 receptors. Finally, it should be pointed

out that we have not considered the possible role of

adenosine A2B receptors in the control of neurotoxicity

since there are no direct evidence for their presence in brain

neurons (see Fredholm et al., 2005), although A2B receptors

are present in other cell types and may indirectly participate

in the control of neurotoxicity by virtue of its ability to

control the expression of cytokines (e.g. Fiebich et al.,

2005).

The present results are surprising since they indicate that

the role of adenosine receptors in the control of neuronal

damage in cultured neurons is fundamentally the opposite of

what is known to occur in adult animals. Thus, whereas A1

receptor activation affords neuroprotection in adult animals

(reviewed by de Mendonca et al., 2000), we now found that

its role in cultured neurons appears to be predominantly

deleterious (see also Turner et al., 2004). In contrast,

whereas blockade of A2A receptors confers a robust brain

neuroprotection in adult animals (reviewed by Cunha,

2005), we now found that it is instead the activation of A2A

receptors that affords neuroprotection in cultured neurons

(see also Ferreira and Paes de Carvalho, 2001). Interestingly,

some studies investigating the role of adenosine receptors in

in vivo models of brain toxicity in newborn animals also

concluded that the role of adenosine receptors was different

from that observed in mature adult animals. For instance, the

group of Rivkees has made a strong case to support that the

tonic activation of A1 receptors has a detrimental role in the

development of the damage in the immature brain, which is

best exemplified by their findings that caffeine and A1

receptor antagonists prevent the prevalent condition of

periventricular leukomalacia in newborns (Turner et al.,

2002a, 2003). Furthermore, some groups described that that

A1 receptor agonists seem to be essentially ineffective in

protecting against ischemia-induced damage in the brain of

newborn animals (Bona et al., 1997; Aden et al., 2001; but

see Hunter et al., 2003). Likewise, ischemia-induced brain

damage in vivo is aggravated in A2A receptor knockout

newborn mice (Aden et al., 2003), the opposite of what is

observed in vivo in adult animals (reviewed by Cunha,

2005), and in essential agreement with the observed

neuroprotective role associated with A2A receptor activation

in cultured neurons.

The effects of endogenous adenosine in the control of

neurotoxicity also seems different in cultured neurons when

compared to the wealth of knowledge gathered in brain

preparations from adult animals subjected to different

insults. In fact, we now observed that endogenous adenosine

saturated A1 receptors (see also Lobner, 2002) and failed to

tonically activate A2A receptors (as well as A3 receptors) in

the control of kainate + cyclothiazide-induced neurotoxi-

city. This contrasts with the marked tonic activation of A2A

receptors in noxious brain conditions in adult animals and to

the tonic activation, but not saturation, of A1 receptors (cf. de

Mendonca et al., 2000; Cunha, 2005). This might be due to a

combined modification of the extracellular metabolism of

adenosine in immature brain preparations, which has been

documented by some studies (see Psarropoulou et al., 1990;

Lynch et al., 1998; Lobner, 2002) but largely remains to be

characterised, and to a different density and sub-cellular

localization of A1 and A2A receptors in immature neurons

and in brain neurons from adult animals. In fact, we now

found that A1 receptor immunoreactivity in cultured cortical

neurons was most evident in the cell body region rather than

on synaptic contacts, whereas A1 receptors are mostly

located in synapses in the adult brain (e.g. Rebola et al.,

2003). Furthermore, A1 receptor expression and density

undergo a striking burst at birth and major increases until P9-

P15 (Rivkees, 1995; Doriat et al., 1999), whereas the

N. Rebola et al. / Neurochemistry International 47 (2005) 317–325324

ontogenic modification of cortical (or extra-striatal) A2A

receptors still remains to be defined.

These modifications of the effects mediated by adenosine

and of the density, sub-cellular localization and tonic

activation of adenosine A1 and A2A receptors should be kept

in mind when considering the use of cultured brain cells as

models to study purinergic modulation in the mature brain.

In fact, it appears that the role of neuronal A1 and A2A

receptors in controlling neuronal viability agrees with their

role in vivo in controlling neurodegeneration in pups/

newborn animals, but does not model the control by

adenosine receptors of neurotoxicity in the brain of adult

animals. Further work is required to determine if these

different effects of adenosine receptors in cultured neurons

(which are similar to these observed in pups/newborn

animals) and in in vivo models of brain toxicity in adult

animals are due to: (1) different density or subcellular

localization of adenosine receptors; (2) different transducing

systems operated by adenosine receptors; (3) different

effects of the transducing systems recruited by adenosine

receptors on neuronal viability during brain development

(see e.g. Ikonomidou et al., 1999; Turner et al., 2002b); (4)

different importance of the control by adenosine receptor in

non-neuronal cells of the development of brain damage in

different brain toxicity models and/or different ages of the

animals.

Acknowledgment

This work was supported by Fundacao para a Ciencia e

Tecnologia (grants SAU/44740/2002 and POCTI/FCB/

36319/1999).

References

Aden, U., Leverin, A.L., Hagberg, H., Fredholm, B.B., 2001. Adenosine A1

receptor agonism in the immature rat brain and heart. Eur. J. Pharmacol.

426, 185–192.

Aden, U., Halldner, L., Lagercrantz, H., Dalmau, I., Ledent, C., Fredholm,

B.B., 2003. Aggravated brain damage after hypoxic ischemia in imma-

ture adenosine A2A knockout mice. Stroke 34, 739–744.

Agostinho, P., Oliveira, C.R., 2003. Involvement of calcineurin in the

neurotoxic effects induced by amyloid-beta and prion peptides. Eur.

J. Neurosci. 17, 1189–196.

Agostinho, P., Caseiro, P., Rego, A.C., Duarte, E.P., Cunha, R.A., Oliveira,

C.R., 2000. Adenosine modulation of D-[3H]aspartate release in cultured

retina cells exposed to oxidative stress. Neurochem. Int. 36, 255–265.

Alfaro, T.M., Vigia, E., Oliveira, C.R., Cunha, R.A., 2004. Effect of free

radicals on adenosine A2A and dopamine D2 receptors in the striatum of

young adult and aged rats. Neurochem. Int. 45, 733–738.

Ambrosio, A.F., Silva, A.P., Malva, J.O., Mesquita, J.F., Carvalho, A.P.,

Carvalho, C.M., 2000. Role of desensitization of AMPA receptors on the

neuronal viability and on the [Ca2+]i changes in cultured rat hippocam-

pal neurons. Eur. J. Neurosci. 12, 2021–2031.

Barth, A., Newell, D.W., Nguyen, L.B., Winn, H.R., Wender, R., Meno, J.R.,

Janigro, D., 1997. Neurotoxicity in organotypic hippocampal slices

mediated by adenosine analogues and nitric oxide. Brain Res. 762, 79–

88.

Bona, E., Aden, U., Gilland, E., Fredholm, B.B., Hagberg, H., 1997.

Neonatal cerebral hypoxia-ischemia: the effect of adenosine receptor

antagonists. Neuropharmacology 36, 1327–1338.

Brambilla, R., Cottini, L., Fumagalli, M., Ceruti, S., Abbracchio, M.P.,

2003. Blockade of A2A adenosine receptors prevents basic fibroblast

growth factor-induced reactive astrogliosis in rat striatal primary astro-

cytes. Glia 43, 190–194.

Brand, A., Vissiennon, Z., Eschke, D., Nieber, K., 2001. Adenosine A1 and

A3 receptors mediate inhibition of synaptic transmission in rat cortical

neurons. Neuropharmacology 40, 85–95.

Costenla, A.R., Lopes, L.V., de Mendonca, A., Ribeiro, J.A., 2001. A

functional role for adenosine A3 receptors: modulation of synaptic

plasticity in the rat hippocampus. Neurosci. Lett. 302, 53–57.

Cunha, R.A., 2005. Neuroprotection by adenosine in the brain: from A1

receptor activation to A2A receptor blockade. Purinergic Signal 1, 111–

134.

Cunha, R.A., Ribeiro, J.A., 2000. Purinergic modulation of [3H]GABA

release from rat hippocampal nerve terminals. Neuropharmacology 39,

1156–1167.

Cunha, R.A., Constantino, M.D., Ribeiro, J.A., 1997. ZM241385 is an

antagonist of the facilitatory responses produced by the A2A adenosine

receptor agonists CGS21680 and HENECA in the rat hippocampus. Br.

J. Pharmacol. 122, 1279–1284.

de Mendonca, A., Sebastiao, A.M., Ribeiro, J.A., 2000. Adenosine: does it

have a neuroprotective role after all? Brain Res. Rev. 33, 258–274.

Diaz-Hernandez, M., Pereira, M.F., Pintor, J., Cunha, R.A., Ribeiro, J.A.,

Miras-Portugal, M.T., 2002. Modulation of the rat hippocampal dinu-

cleotide receptor by adenosine receptor activation. J. Pharmacol. Exp.

Ther. 301, 441–450.

Doriat, J.F., Koziel, V., Humbert, A.C., Daval, J.L., 1999. Medium- and

long-term alterations of brain A1 and A2A adenosine receptor charac-

teristics following repeated seizures in developing rats. Epilepsy Res.

35, 219–228.

Fedorova, I.M., Jacobson, M.A., Basile, A., Jacobson, K.A., 2003. Beha-

vioral characterization of mice lacking the A3 adenosine receptor:

sensitivity to hypoxic neurodegeneration. Cell. Mol. Neurobiol. 23,

431–447.

Ferreira, J.M., Paes de Carvalho, R., 2001. Long-term activation of ade-

nosine A2A receptors blocks glutamate excitotoxicity in cultures of

avian retinal neurons. Brain Res. 900, 169–176.

Fiebich, B.L., Akundi, R.S., Biber, K., Hamke, M., Schmidt, C., Butcher,

R.D., van Calker, D., Willmroth, F., 2005. IL-6 expression induced by

adenosine A2b receptor stimulation in U373 MG cells depends on p38

mitogen activated kinase and protein kinase C. Neurochem. Int. 46,

501–512.

Fredholm, B.B., 1997. Adenosine and neuroprotection. Int. Rev. Neurobiol.

40, 259–280.

Fredholm, B.B., Chen, J.F., Cunha, R.A., Svenningsson, P., Vaugeois, J.M.,

2005. Adenosine and brain function. Int. Rev. Neurobiol. 63, 191–270.

Hammarberg, C., Schulte, G., Fredholm, B.B., 2003. Evidence for func-

tional adenosine A3 receptors in microglia cells. J. Neurochem. 86,

1051–1054.

Hunter, C.J., Bennet, L., Power, G.G., Roelfsema, V., Blood, A.B., Quae-

dackers, J.S., George, S., Guan, J., Gunn, A.J., 2003. Key neuropro-

tective role for endogenous adenosine A1 receptor activation during

asphyxia in the fetal sheep. Stroke 34, 2240–2245.

Ikonomidou, C., Bosch, F., Miksa, M., Bittigau, P., Vockler, J., Dikranian,

K., Tenkova, T.I., Stefovska, V., Turski, L., Olney, J.W., 1999. Blockade

of NMDA receptors and apoptotic neurodegeneration in the developing

brain. Science 283, 70–74.

Lee, H.K., Choi, S.S., Han, K.J., Han, E.J., Suh, H.W., 2004. Roles of

adenosine receptors in the regulation of kainic acid-induced neurotoxic

responses in mice. Mol. Brain Res. 125, 76–85.

Lobner, D., 2002. Saturation of neuroprotective effects of adenosine in

cortical culture. Neuroreport 13, 2075–2078.

N. Rebola et al. / Neurochemistry International 47 (2005) 317–325 325

Lobner, D., Choi, D.W., 1994. Dipyridamole increases oxygen–glucose

deprivation-induced injury in cortical cell culture. Stroke 25, 2085–

2089.

Logan, M., Sweeney, M.I., 1997. Adenosine A1 receptor activation pre-

ferentially protects cultured cerebellar neurons versus astrocytes against

hypoxia-induced death. Mol. Chem. Neuropathol. 31, 119–133.

Lopes, L.V., Rebola, N., Pinheiro, P.C., Richardson, P.J., Oliveira, C.R.,

Cunha, R.A., 2003a. Adenosine A3 receptors are located in neurons of

the rat hippocampus. Neuroreport 14, 1645–1648.

Lopes, L.V., Rebola, N., Costenla, A.R., Halldner, L., Jacobson, M.A.,

Oliveira, C.R., Richardson, P.J., Fredholm, B.B., Ribeiro, J.A., Cunha,

R.A., 2003b. Adenosine A3 receptors in the rat hippocampus: lack of

interaction with A1 receptors. Drug Dev. Res. 58, 428–438.

Lopes, L.V., Halldner, L., Rebola, N., Johansson, B., Ledent, C., Chen, J.F.,

Fredholm, B.B., Cunha, R.A., 2004. Binding of the prototypical ade-

nosine A2A receptor agonist CGS 21680 to the cerebral cortex of

adenosine A1 and A2A receptor knockout mice. Br. J. Pharmacol.

141, 1006–1014.

Lupica, C.R., Cass, W.A., Zahniser, N.R., Dunwiddie, T.V., 1990. Effects

of the selective adenosine A2 receptor agonist CGS 21680 on in

vitro electrophysiology, cAMP formation and dopamine release in

rat hippocampus and striatum. J. Pharmacol. Exp. Ther. 252, 1134–

1141.

Lynch, J.J., Alexander, K.M., Jarvis, M.F., Kowaluk, E.A., 1998. Inhibition

of adenosine kinase during oxygen–glucose deprivation in rat cortical

neuronal cultures. Neurosci. Lett. 252, 207–210.

Psarropoulou, C., Kostopulos, G., Haas, H.L., 1990. An electrophysiolo-

gical study of the ontogenesis of adenosine receptors in the CA1 area of

rat hippocampus. Dev. Brain Res. 55, 147–150.

Pugliese, A.M., Latini, S., Corradetti, R., Pedata, F., 2003. Brief, repeated,

oxygen–glucose deprivation episodes protect neurotransmission from a

longer ischemic episode in the in vitro hippocampus: role of adenosine

receptors. Br. J. Pharmacol. 140, 305–314.

Rebola, N., Pinheiro, P.C., Oliveira, C.R., Malva, J.O., Cunha, R.A., 2003.

Subcellular localization of adenosine A1 receptors in nerve terminals

and synapses of the rat hippocampus.. Brain Res. 987, 49–58.

Rebola, N., Canas, P., Oliveira, C.R., Cunha, R.A., 2005. Different synaptic

and subsynaptic localization of adenosine A2A receptors in the hippo-

campus and striatum of the rat. Neuroscience 132, 893–903.

Rivkees, S.A., 1995. The ontogeny of cardiac and neural A1 adenosine

receptor expression in rats. Dev. Brain Res. 89, 202–213.

Rivkees, S.A., Thevananther, S., Hao, H., 2001. Are A3 adenosine receptors

expressed in the brain? Neuroreport 11, 1025–1030.

Scholz, K.P., Miller, R.J., 1991. Analysis of adenosine actions on Ca2+

currents and synaptic transmission in cultured rat hippocampal pyra-

midal neurones. J. Physiol. 435, 373–393.

Turner, C.P., Yan, H., Schwartz, M., Othman, T., Rivkees, S.A., 2002a. A1

adenosine receptor activation induces ventriculomegaly and white

matter loss. Neuroreport 13, 1199–1204.

Turner, C.P., Pulciani, D., Rivkees, S.A., 2002b. Reduction in intracellular

calcium levels induces injury in developing neurons. Exp. Neurol. 178,

21–32.

Turner, C.P., Seli, M., Ment, L., Stewart, W., Yan, H., Johansson, B.,

Fredholm, B.B., Blackburn, M., Rivkees, S.A., 2003. A1 adenosine

receptors mediate hypoxia-induced ventriculomegaly. Proc. Natl. Acad.

Sic. U.S.A. 100, 11718–11722.

Turner, C.P., Blackburn, M.R., Rivkees, S.A., 2004. A1 adenosine receptors

mediate hypoglycemia-induced neuronal injury. J. Mol. Endocrinol. 32,

129–144.

von Lubitz, D.K., Lin, R.C., Popik, P., Carter, M.F., Jacobson, K.A., 1994.

Adenosine A3 receptor stimulation and cerebral ischemia. Eur. J.

Pharmacol. 263, 59–67.