Address correspondence and reprint requests to Josep Saura, Department Pharmacology and Toxicology, IIBB, CSIC-IDIBAPS, Rossello 161, planta 6, 08036-Barcelona, Spain. E-mail: firstname.lastname@example.org
The absence of adenosine A2A receptors, or its pharmacological inhibition, has neuroprotective effects. Experimental data suggest that glial A2A receptors participate in neurodegeneration induced by A2A receptor stimulation. In this study we have investigated the effects of A2A receptor stimulation on control and activated glial cells. Mouse cortical mixed glial cultures (75% astrocytes, 25% microglia) were treated with the A2A receptor agonist CGS21680 alone or in combination with lipopolysaccharide (LPS). CGS21680 potentiated lipopolysaccharide-induced NO release and NO synthase-II expression in a time- and concentration-dependent manner. CGS21680 potentiation of lipopolysaccharide-induced NO release was suppressed by the A2A receptor antagonist ZM-241385 and did not occur on mixed glial cultures from A2A receptor-deficient mice. In mixed glial cultures treated with LPS + CGS21680, the NO synthase-II inhibitor 1400W abolished NO production, and NO synthase-II immunoreactivity was observed only in microglia. Binding experiments demonstrated the presence of A2A receptors on microglial but not on astroglial cultures. However, the presence of astrocytes was necessary for CGS21680 potentiating effect. In light of the reported neurotoxicity of microglial NO synthase-II and the neuroprotection of A2A receptor inhibition, these data suggest that attenuation of microglial NO production could contribute to the neuroprotection afforded by A2A receptor antagonists.
The mechanism for the neuroprotection afforded by A2AR antagonists has not yet been elucidated. Several hypotheses have been postulated (see Schwarzschild et al. 2003 for review) but some of them only explain such neuroprotective effects on Parkinson's disease models. A hypothesis that could be valid for all the in vivo models reported so far maintains that A2AR stimulation modulates astroglial and/or microglial cell function resulting in deleterious effects for surrounding neurons. Two ways have been proposed by which A2AR-stimulated glial cells could be harmful to neurons. On one hand, stimulation of A2AR on astroglial cultures increases glutamate release that can be inhibited by A2AR antagonists (Li et al. 2001; Nishizaki et al. 2002). The resulting increase in extracellular glutamate levels could induce excitotoxic neuronal damage. On the other hand, stimulation of A2AR on microglial cells in culture results in an activated phenotype (Fiebich et al. 1996; Gebicke-Haerter et al. 1996). Activated microglial cells have been shown to release a number of potentially neurotoxic factors (Streit 2004). Especially compelling is the evidence showing the neurotoxic effects of NO produced by microglial NO synthase-II (NOS-II) (Liberatore et al. 1999; Dehmer et al. 2000; Bal-Price and Brown 2001; Golde et al. 2002).
Recently, we have observed that microglial cells show A2AR immunoreactivity in cortical samples from Alzheimer disease patients but not in controls (Angulo et al. 2003). Other authors have shown that the microglial activator lipopolysaccharide (LPS) increases A2AR mRNA (Wittendorp et al. 2004) and protein (Canas et al. 2004) in microglial cells in culture. These data show that activated microglia have increased levels of A2AR and therefore activated microglia could be especially sensitive to A2AR agonists. To verify this hypothesis, we have undertaken a study in which the effects of A2AR stimulation on NO release have been compared on control and activated microglia. We have used mixed glial cultures as a model because this system permits the cross-talk between astrocytes and microglia.
Materials and methods
Trypsin-EDTA solution (no. 25200-072), Dulbecco's modified Eagle's medium-F-12 nutrient mixture (no. 31330-038), fetal bovine serum (no. 10270-106) and penicillin–streptomycin (no. 15140-114) were from Invitrogen (Carlsbad, CA, USA). Deoxyribonuclease I (no. D-5025), biotin-labelled tomato lectin (no. L-9389), extravidin peroxidase (no. E-2886), bisbenzimide (Hoechst 33258, no. B-1115), laminin (no. L-2020), U0126 (no. U-120), H89 (no. B-1427), cytosine arabinoside (no. C-1768), rabbit anti-actin (no. A-5060) and Escherichia coli lipopolysaccharide (LPS, no. L-2654) were from Sigma (St Louis, MO, USA). Rabbit anti-cow glial fibrillary acidic protein (GFAP) (no. Z0334) was from Dako (Carpinteria, CA, USA). Alexa Fluor 488 goat anti-mouse (no. A11017) and Alexa Fluor 546 (no. 11010) goat anti-rabbit antibodies were from Molecular Probes (Eugene, OR, USA). Fluorescein-conjugated streptavidin (no. SA103) and rabbit anti-NOS-II antibody (no. AB16311) were from Chemicon (Temecula, CA, USA). TNF-α enzyme-linked immunosorbent assay (ELISA) kit (no. 87.010.010) was from Diaclone (Besançon, France). 1400W (no. 1415), CGS21680 (no. 1063), GF109203X (no. 0741), ZM-241385 (no. 1036) and [3H]ZM-241385 (no. R1036) were from Tocris (Avonmouth, UK). Mouse anti-NOS-II was from BD Transduction Laboratories (Lexington, KY, USA). Polyvinylidene difluoride membranes (no. IPVH00010) were from Millipore (Bedford, MA, USA). HRP-labelled anti-rabbit antibody (no. NA934) and ECL-Plus (no. RPN2132) were from Amersham (Buckinghamshire, UK). X-ray film (no. RP2) was from Agfa (Mortsel, Belgium). Glass filters (GF/C filters) were from Whatman (Kent, UK). Scintillation solution EcoscintH was from National Diagnostics (Atlanta, GA, USA).
C57BL/6 wild-type mice (Charles River, Lyon, France) were used in most experiments. A2AR-deficient mice were also used alongside their littermate wild-type control mice. Mice with genetically disrupted A2AR were generated using standard gene targeting by DNA recombination on a mixed 129-Steel × C57BL/6 genetic background (Chen et al. 1999). The congenic A2AR knockout mice in C57BL/6 background were achieved by back-crossing the A2AR knockout in mixed (129-Steel × C57BL/6) genetic background to C57BL/6 mice for more than six generations. Genomic DNA was isolated from mouse tails, and the genotype of mouse was determined by PCR analysis as described previously (Yu et al. 2004). Animals were kept in the animal house at the Cajal Institute and were housed between four and six animals per cage and maintained with free access to food and water, under a 12 h light/dark cycle, at room temperature (22 ± 2°C), according to standard guidelines approved by the local ethical committee, following European Community Guidelines on the Care and Use of Laboratory Animals.
Mouse mixed glial cultures were prepared as previously described (Giulian and Baker 1986). Cerebral cortices from 1- or 2-day-old neonatal mice were dissected, carefully stripped of their meninges and digested with 0.25% trypsin for 30 min at 37°C. Trypsinization was stopped by adding an equal volume of culture medium (Dulbecco's modified Eagle's medium-F-12 nutrient mixture, fetal bovine serum 10%, penicillin 100 U/mL and streptomycin 100 μg/mL) to which 0.02% deoxyribonuclease I was added. The solution was pelleted, re-suspended in culture medium and brought to a single cell suspension by repeated pipetting followed by passage through a 105-μm pore mesh. Cells were seeded at a density of 250 000 cells/mL (= 62 500 cells/cm2) and cultured at 37°C in humidified 5% CO2−95% air. Medium was replaced every 4–5 days. Cultures reached confluency after 7–10 days in vitro and were used between 12 and 20 days in vitro. At this point they typically consisted of 75% type-I astrocytes and 25% microglia.
Mouse microglial cultures were prepared by mild trypsinization as described (Saura et al. 2003). Briefly, confluent mixed glial cultures were subjected to mild trypsinization (0.06%) in the presence of 0.25 mm EDTA and 0.5 mm Ca2+. This results in the detachment of an intact layer of cells containing virtually all the astrocytes and leaves a population of firmly attached cells identified as > 98% microglia. Twenty-four hours after isolation by this procedure microglial cultures were used.
Mouse astroglial cultures were prepared by seeding cortical cells prepared as for mixed glial cultures on laminin (20 μg/mL)-coated plates. Laminin-coating favours astroglial growth and inhibits microglial growth (Milner and Campbell 2002). At confluency cultures were treated with 10 μm cytosine arabinoside for 4 days. Cultures were used 1 day after the end of cytosine arabinoside treatment. The vast majority of cells were type-I astrocytes and microglia were less than 2%.
Drugs were prepared in culture medium as 15-fold concentrated stocks. Vehicle-containing solution was added to control wells. For LPS and CGS21680 co-treatment, CGS21680 was added 15 min before LPS. ZM-241385, H89, GF109203X and U0126 were added 30 min before CGS21680. 1400W was added 15 min after LPS.
NO production was assessed by the Griess reaction, a colorimetric assay that detects nitrite (NO2–), a stable reaction product of NO and molecular oxygen, in culture supernatants. Briefly, 50 μL of each sample was incubated with 100 μL of Griess reagent A (1% sulfanilamide, 5% phosphoric acid) for 5 min followed by addition of 100 μL of Griess reagent B (0.1% N-1-naphtylenediamine) for 5 min. The optical density of the samples was measured at 540 nm using a microplate reader (iEMS Reader MF, Labsystems, Barcelona, Spain). The nitrite concentration was determined from a sodium nitrite standard curve.
The amount of TNF-α released in the cell culture supernatants (100 μL medium conditioned for 6 h) was determined using an ELISA kit and following the instructions supplied by the manufacturer.
Twenty micrograms of protein of denatured (5 min, 100°C) mixed glial total extracts were subjected to sodium dodecyl sulfate – polyacrylamide gel electrophoresis on a 7% polyacrylamide gel and transferred to a polyvinylidene difluoride membrane. The membranes were processed as follows: (i) 5 min in Tris-buffered saline (TBS, 20 mm Tris, 0.15 m NaCl, pH = 7.5); (ii) 1 min in methanol; (iii) air dry; (iv) overnight incubation at 4°C on primary antibody (rabbit anti-NOS-II 1 : 200 or rabbit anti-actin 1 : 1000) diluted in immunoblot buffer (TBS containing 0.05% Tween-20 and 5% non-fat dry milk); (v) 2 × 15 s in TBS 0.05% Tween-20; (vi) 1 h in HRP-labelled anti-rabbit antibody diluted 1 : 5000 in immunoblot buffer; (vii) extensive washes in TBS-0.05% Tween-20; (viii) 5 min in ECL-Plus. Membranes were apposed to an X-ray film that was developed and quantified using a scanning densitometer.
Immunocytochemistry and lectin staining
Forty-eight hours after treatment mixed glial cultures were fixed with 4% paraformaldehyde (60 min, room temperature) and stained with tomato lectin as follows; (i) 15 min in phosphate-buffered saline (PBS); (ii) 5 min in 0.3% H2O2 in methanol; (iii) 2 × 10 min in PBS; (iv) overnight incubation at 4°C in tomato lectin diluted 1 : 500 in PBS containing 7% normal goat serum; (v) 2 × 10 min in PBS; (vi) 1 h in HRP-labelled extravidin diluted 1 : 500 in PBS containing 7% normal goat serum; (vii) 2 × 10 min in PBS; (viii) 5 min in PBS containing diaminobenzidine (1 μg/mL) and H2O2 (1.5 μL/mL). After lectin staining, 10 fields of 0.148 mm2 were photographed per well with a Nikon DXM1200 camera under the × 20 objective. The photographed fields were selected using a predetermined grid and an investigator unaware of the treatment counted tomato lectin-positive cells.
To identify NO-producing cells in mixed glial cultures, double immunocytochemistry was performed using specific markers for each glial cell type and a specific monoclonal antibody for NOS-II, which is responsible for the synthesis of NO. Rabbit anti-GFAP polyclonal antibody and biotin-labelled tomato lectin were used to identify astroglial and microglial cells, respectively. Cells grown on glass coverslips were fixed, permeated and non-specific staining was blocked by incubating the cells with 5% normal goat serum in PBS/1% bovine serum albumin for 20 min at room temperature. Cells were co-incubated overnight at 4°C with either rabbit anti-GFAP polyclonal antibody (1/100) and anti-NOS-II monoclonal antibody (1/500) or tomato lectin (1/400) and anti-NOS-II monoclonal antibody (1/500). After rinsing in PBS, cells were co-incubated for 1 h at room temperature with fluorescent secondary antibodies: goat anti-rabbit Alexa 488 (1/1000) for the rabbit anti-GFAP polyclonal antibody and Alexa 546 goat anti-mouse antibody (1/1000) for the NOS-II monoclonal antibody or FITC streptavidin (1/400) for tomato lectin, and Alexa 546 goat anti-mouse antibody (1/1000) for the NOS-II monoclonal antibody. After rinsing in PBS, cells were incubated for 5 min with 100 nm of Hoechst 33258. Samples were visualized under a fluorescence microscope (Nikon Eclipse E1000).
Radioligand binding assays
For binding assays untreated or LPS-treated cells (astroglia, microglia or mixed cultures) were re-suspended in 50 mm Tris-HCl buffer, pH 7.4 and homogenized using a Polytron (Kinematica, three strokes of 5 s each one). After 30 min on ice, the homogenate was centrifuged at 105 000 g for 45 min (4°C). The resulting membrane pellet was re-suspended in 50 mm Tris-HCl buffer and centrifuged under the same conditions. Membranes were re-suspended in 50 mm Tris-HCl buffer, pH 7.4 containing 10 mm MgCl2 and 2 UI/mL adenosine deaminase and radioligand binding was determined by the addition of 2 nm[3H]ZM-241385 in the absence or presence (non-specific binding) of 10 μm ZM-241385 or 50 μm CGS21680. When equilibrium had been reached (60 min at room temperature), incubates were filtered through glass fibre filters in a Brandel cell harvester (Biomedical Research and Development Laboratories, Gaithersburg, MD, USA). Filters were washed in 5 mL of ice-cold Tris-HCl buffer and transferred to vials containing 10 mL of scintillation solution. After overnight shaking, vials were counted using a Packard 1600 TRI-CARB scintillation counter with 60% efficiency. All points represent the mean of triplicates.
The statistical significance of the difference between means was performed with Student t-test (two samples) or anova followed by Bonferroni's post-hoc test (more than two samples). SPSS 11.5 (SPSS Inc., Chicago, IL, USA) was used. For EC50 estimation Prism 4 (GraphPad Software Inc, San Diego, CA, USA) was used. Graphic data are expressed as mean + SEM.
CGS21680 potentiates LPS-induced NO release
The ability of the specific A2AR agonist CGS21680 to induce NO release was studied in mouse mixed glial cultures. CGS21680 did not induce NO release when given alone but, in contrast, the reagent potentiated LPS-induced NO release in a concentration-dependent manner with a maximal potentiation at 100 nm(Fig. 1a) and an estimated EC50 of 27 nm. Kinetic studies revealed that this effect was time-dependent and could be observed at 36 and 48 h but not before (Fig. 1b). We also analyzed the effects of CGS21680 on LPS-induced release of the pro-inflammatory cytokine TNF-α. Unlike NO release, LPS-induced release of TNF-α was not potentiated by CGS21680 (Fig. 2).
CGS21680 potentiation of LPS-induced NO release is mediated by A2AR
To demonstrate that the potentiating effect of CGS21680 on LPS-induced NO release was indeed mediated by A2AR, the selective antagonist ZM-241385 was used. As expected, 5 μm of ZM-241385 completely abolished the CGS21680 potentiation of NO release but it did not affect LPS-induced NO release in the absence of CGS21680 (Fig. 3). The involvement of A2AR was confirmed by using mixed glial cultures from A2AR deficient mice. As shown in Fig. 4, CGS21680 failed to potentiate LPS-induced NO release in mixed glial cultures lacking the A2AR.
NO released by LPS + CGS21680 in mixed glial cultures is produced by microglial NOS-II
To determine whether type II NO synthase (NOS-II) was responsible for the CGS21680-induced NO release potentiation, mixed glial cultures were treated with the selective NOS-II inhibitor 1400W. In the presence of 30 μm 1400W, neither LPS nor LPS + CGS21680 induced NO release (Fig. 5a), indicating that in both cases NO is produced exclusively by NOS-II. This finding was supported by the observation that CGS21680 also potentiates the LPS-induced increase in NOS-II protein levels (Fig. 5b).
All above experiments had been performed on mixed glial cultures, which basically comprise two cell types: astrocytes and microglia. To determine the cell type producing CGS21680-induced NO release potentiation, the cellular localization of NOS-II in mixed glial cultures treated with LPS + CGS21680 was analyzed. By double immunofluorescence it was observed that NOS-II immunostaining always co-localized with the microglial marker tomato lectin (Fig. 6a). In contrast, NOS-II immunoreactivity was never observed in GFAP-labelled astrocytes (Fig. 6b). These results demonstrate that the CGS21680-induced effect in mixed glial cultures is mediated by microglial NOS-II.
CGS21680 potentiation of LPS-induced NO release is not caused by increased number of microglial cells
Previous reports have shown that adenosine, or a combination of A1R and A2AR agonists, induce microglial proliferation (Rathbone et al. 1992; Gebicke-Haerter et al. 1996), although in other studies a depressed proliferation rate has been observed (Schubert et al. 1997; Si et al. 1996). We therefore considered the possibility that the CGS21680-induced effects on microglial NOS-II levels and NO release were as a result of an increase in the number of microglial cells caused by an enhanced proliferation. To test this hypothesis, mixed glial cultures were treated with LPS, LPS + CGS21680, CGS21680 or vehicle for 48 h and microglial cells were stained with tomato lectin and counted. As shown in Fig. 7, the number of microglial cells was not higher in cultures treated with LPS + CGS21680 (100 nm or 1 μm) when compared with LPS-treated cultures. This result demonstrates that CGS21680 potentiation of LPS-induced NO release is not because of an increase in the number of microglial cells.
CGS21680 does not potentiate LPS-induced NO release in microglial-enriched cultures
The effect of CGS21680 on LPS-induced NO release was determined in isolated microglial cultures. Microglial cultures were treated with LPS ± CGS21680 and NO release at 48 h was compared with that obtained in mixed glial cultures. Interestingly, CGS21680 did not potentiate LPS-induced NO production by microglial cultures (Fig. 8). This finding, together with the lack of NOS-II in astrocytes (Fig. 6), shows that the presence of astrocytes is necessary for CGS21680 to induce its potentiating effect on microglial NO release.
A2AR expression in vitro is detectable in murine microglia and not in astrocytes
At this point, a key question was whether A2AR are present on mouse microglia, astrocytes, or both, in vitro. We initially addressed this question by immunocytochemistry using an antibody that specifically recognizes the A2A receptor in sections of mouse striatum. Unfortunately, we could not observe any signal in mixed glial cultures that was above the signal obtained in cultures prepared from A2A-deficient mice (not shown). Because we presumed that the levels of A2AR on glial cultures could be below the sensitivity threshold for the technique, we addressed the question by using a more sensitive approach, which is [3H]ZM-241385 binding to membranes from glial cultures. According to the adenosine receptor pharmacology (Poucher et al. 1995; Alexander and Millns 2001), binding of 2 nm[3H]ZM-241385 is specific for A2AR (A2AR saturation higher than 80%; A1R, A2BR or A3R saturation less than 2%). As shown in Fig. 9, microglial cells, but not astroglial cells, appreciably bound the A2AR-selective antagonist. In mixed glial cultures, the specific binding was lower than in microglial cultures in agreement with the low percentage of microglia in these primary cultures (Fig. 9). LPS treatment induced a significant increase (35%) in [3H]ZM-241385 binding to membranes from microglial cultures and not from mixed glial or astroglial cultures (Fig. 9). In all cases, [3H]ZM-241385 binding to mouse striatal membranes (824 ± 13 fmol/mg prot) was at least 10 times higher than binding to membranes from glial cultures.
Signal transduction involved in CGS21680 potentiation of LPS-induced NO release
The classical signal transduction triggered by A2AR activation includes intracellular cAMP production and ERK1/2 (extracellular signal-regulated kinases) phosphorylation (Haas and Selbach 2000; Ferréet al. 2002; Canals et al. 2005). It is also known that protein kinase C (PKC) activation mediates LPS-induced NO release (Fiebich et al. 1998; Nakamura et al. 2001; Sunohara et al. 2001; Nakajima et al. 2003). We therefore investigated the involvement of these signal transduction pathways in the CGS21680 potentiation of LPS-induced NO release in mixed glial cultures. On one hand, the PKA inhibitor H89 (10 μm) increased LPS-induced NO release, which is in agreement with the reported inhibition of LPS-induced NO release by protein kinase A (PKA) (Fiebich et al. 1998; Kim et al. 2000, 2004), but it did not suppress the potentiating effect of CGS21680. On the other hand, the CGS21680-mediated potentiation did not occur in the presence of the PKC inhibitor GF109203X (5 μm) or the MEK inhibitor U0126 (10 μm) (Fig. 10), suggesting that CGS21680 potentiation of LPS-induced NO release depends on signaling molecules upstream PKC and ERK1/2.
In this study we show for the first time that the A2AR agonist CGS21680, which by itself has no effect on glial NO release, markedly potentiates the effect of LPS on NO release by mixed glial cultures. Yet, TNF-α release, which is strongly induced by LPS, is not potentiated by CGS21680, indicating that not every effect of LPS on glial cells is enhanced by CGS21680. It should be noted that, because of their different kinetics, TNFα and NO could not be measured at the same time after treatment and were measured at 6 and 48 h, respectively.
We have clearly demonstrated that CGS21680 potentiation of LPS-induced NO release is mediated by A2AR. On one hand, the CGS21680 concentration–response of glial NO release in the presence of LPS shows an EC50 of 27 nm, which fits with the KD values (20–60 nm) reported for this ligand (Klotz et al. 1998) and a maximal potentiation at a concentration, 100 nm, at which CGS21680 is a selective A2AR agonist (Klotz et al. 1998). On the other hand, CGS21680 potentiation of LPS-induced NO release is completely prevented by the A2AR antagonist ZM-241385. At the concentration used of 5 μm, ZM-241385 acts on both A2AR and A2BR. As it would be unexpected for concentrations of CGS21680 below 10 μm to act at A2BR, and we used 100 nm in this study, we assume that ZM-241385 antagonizes the potentiating effect of CGS21680 by blocking A2AR. Finally, CGS21680 failed to induce the potentiation effect in mixed glial cultures from mice lacking A2AR.
We have also demonstrated that microglial NOS-II is responsible for NO production induced by LPS + CGS21680 in mixed glial cultures. First, we have shown that NO release induced by LPS + CGS21680 is fully blocked by 1400W, a selective NOS-II inhibitor. In agreement with this, we have observed that CGS21680 potentiates LPS-induced increase in NOS-II protein levels. Then, by double immunofluorescence in mixed glial cultures treated with LPS + CGS21680, we have localized NOS-II protein in microglial cells and not in astrocytes. These results clearly indicate that A2AR stimulation potentiates LPS-induced release of NO produced by microglial NOS-II. Our finding that CGS21680 potentiates LPS-induced microglial NO production contrasts with the results of Brodie et al. (1998), who observed that CGS21680 inhibits NO production induced by LPS + γ-interferon in C6 cells. The concentration of CGS21680 needed for such an effect was higher than those used in this study, with no effects observed at 250 nm, suggesting that targets other than A2AR may participate in that response. More importantly, that study was performed on the rat C6 glioma cell line, which has a mixed astroglial–oligodendroglial phenotype (Coyle 1995). These cells are from a different lineage than microglia, have a different function and may therefore respond differently.
Adenosine has been shown to inhibit activation of immune cells through A2AR (Sitkovsky et al. 2004). The physiological significance of this effect is thought to be to protect tissue from acute inflammatory damage. Therefore, A2AR agonists have potential clinical application as anti-inflammatory agents. By using a chimeric animal model consisting of wild-type mice transplanted with A2AR knockout bone marrow cells, Yu et al. (2004) have recently demonstrated that selective inactivation of A2AR on bone marrow derived cells exacerbates ischaemic liver injury, but attenuates infarct volumes and ischaemia-induced expression of pro-inflammatory cytokines in the brain. Because NO production by NOS-II is a typical inflammatory response, our results confirm an opposite regulation of inflammation by adenosine in brain compared with peripheral organs. Previous reports have shown that stimulation of A2AR in microglia results in responses that reflect an activated phenotype: increased proliferation (Gebicke-Haerter et al. 1996), increased expression of cyclooxygenase-2, prostaglandin-E2 (Fiebich et al. 1996), nerve growth factor (Heese et al. 1997) and Kv1.3 (Kust et al. 1999). Altogether, these findings seem to indicate that the microglial response to A2AR stimulation is pro-inflammatory, unlike that of most immune cells, and that adenosine antagonists would be anti-inflammatory and protective in brain but pro-inflammatory and exacerbating damage in liver. This differential response will have to be taken into account when designing A2AR agonists as peripheral anti-inflammatory agents and suggests a potential advantage of A2AR agonists with low CNS penetration. However, antagonists targeting A2AR in microglia may be useful in neurological diseases with an inflammatory component.
Several studies have investigated the effects of A2AR stimulation on glial cells. The present study includes some novel elements, one of them being the use of mixed glial cultures. Unlike previous studies using astroglial- or microglial-enriched cultures, we have used mixed glial cultures comprising approximately 75% astrocytes and 25% microglia. These preparations permit the cross-talk between both cell types, which is of great importance in the process of glial activation. Our findings prove the value of using mixed glial cultures. Indeed, the potentiation elicited by CGS21680 on LPS-induced NO release by microglial cells only occurs when astrocytes are present. Evidence for the influence of astrocytes on LPS-induced microglial NO release has been reported previously. Thus, we have shown that the presence of astrocytes markedly potentiates LPS-induced microglial NO release, and that factors present in the plasma membrane contribute to this effect (Solàet al. 2002). Also, in a recent study, Pyo et al. (2003) have shown that wortmannin, probably not through PI-3 kinase inhibition, potentiates LPS-induced microglial NO release only in the presence of astrocytes. An explanation for the requirement of astrocytes for CGS21680 potentiation effect would be that astrocytes are the primary target for CGS21680 on mixed glial cultures. However, our data from binding experiments do not support this possibility as we have observed A2AR in microglial but not in astroglial cultures. In agreement with this, Wittendorp et al. (2004), using PCR in mouse cultures, found A2AR in activated microglia but not in astrocytes and Peakman and Hill (1996) failed to observe functional A2AR on cultured type-I rat astrocytes. Other groups have observed functional effects of A2AR agonists or antagonists on rat astroglial cultures, suggesting that rat astrocytes are capable of expressing A2AR (Li et al. 2001; Nishizaki et al. 2002; Brambilla et al. 2003). A species-related difference or the presence of contaminating microglia in astroglial-enriched cultures are possible explanations for this discrepancy. In contrast, CGS21680 does not potentiate LPS-induced NO release by microglia, even when microglial cells are cultured with medium conditioned by astrocytes or mixed glia (data not shown). This suggests that astrocytes but not soluble factors released from astrocytes are required for A2AR potentiation of LPS-induced NO release by microglia. The involvement of a soluble factor cannot be completely ruled out as the reported lack of effect of conditioned medium can be attributed to the instability of the soluble factor(s) or the unsuitability of the time frame used for medium conditioning.
Another novel element in this study is that we have investigated the effects of CGS21680 on glial cells, not only when given alone, but also in combination with LPS, which activates glial cells, particularly microglia. In contrast to previous studies reporting activation of microglial cells by CGS21680 (Gebicke-Haerter et al. 1996; Fiebich et al. 1996; Heese et al. 1997; Kust et al. 1999), and despite the fact that A2AR are expressed in non-activated microglial cells, we did not observe any NO release induced by CGS21680 alone. However, a marked potentiation of NO release was observed if cells were also treated with LPS. Activated microglia therefore appear to be especially sensitive to A2AR stimulation. The slight (35%) up-regulation of the receptor in LPS-activated cells, which is in agreement with the results of Wittendorp et al. (2004) and Canas et al. (2004), who showed LPS-induced increases in A2AR mRNA and immunoreactivity, respectively, cannot account for the effect. Thus, these findings indicate that A2AR agonists regulate NO release in microglia by a signalling pathway that is operative in activated but not in resting cells. Interestingly, LPS synergizes with A2AR agonists to up-regulate the expression of vascular endothelial growth factor, but not TNF-α, in mouse macrophages (Leibovich et al. 2002; Pinhal-Enfield et al. 2003). This interaction has been proposed as a mechanism that directs the macrophage towards an angiogenic phenotype.
The few studies on A2AR expression in human microglia indicate that the receptor is preferentially expressed in activated microglia, i.e. when microglia responds to a pathological challenge (Angulo et al. 2003). With the necessary caution when extrapolating data obtained using primary cultures, our findings offer an interesting link between the neurotoxic effects of activated microglial NOS-II (see Introduction for references) and the neuroprotective effects of A2AR antagonists (see Introduction for references). In an ongoing neuropathological situation, NO-induced toxicity caused by activated microglia could be exacerbated by extracellular adenosine – whose levels increase by neuronal damage – acting on microglial A2A receptors. This opens new perspectives for pharmacological intervention in neuroinflammatory diseases.
We thank Dr M. C. Montesinos for help on PCR experiments for genotyping. This work was supported by grants from Spanish Ministerio de Ciencia y Tecnología (SAF2001-2240 and Red CIEN to JS, SAF2002-03293 to RF and SAF2003-4864; GEN2003-20651-C06-02 to RM); from Spanish Fondo de Investigaciones Sanitarias (PI040778 to JS; RTA-G03/05 to RM); from Fundació Marató TV3 (01/012710 to RF) and Fundació la Caixa (2003 × 928 to RM and 02/056–00 to RF); from National Institute of Health (ES10804) to MS. AE is recipient of a pre-doctoral grant from IDIBAPS.