Neuroprotection by caffeine and adenosine A2A receptor blockade of β-amyloid neurotoxicity


Center for Neuroscience of Coimbra, Institute of Biochemistry, Faculty of Medicine, University of Coimbra, 3004-504 Coimbra, Portugal; E-mail:


Adenosine is a neuromodulator in the nervous system and it has recently been observed that pharmacological blockade or gene disruption of adenosine A2A receptors confers neuroprotection under different neurotoxic situations in the brain. We now observed that coapplication of either caffeine (1–25 μM) or the selective A2A receptor antagonist, 4-(2-[7-amino-2(2-furyl)(1,2,4)triazolo (2,3-a)(1,3,5)triazin-5-ylamino]ethyl)phenol (ZM 241385, 50 nM), but not the A receptor antagonist, 8-cyclopentyltheophylline (200 nM), prevented the neuronal cell death caused by exposure of rat cultured cerebellar granule neurons to fragment 25–35 of β-amyloid protein (25 μM for 48 h), that by itself caused a near three-fold increase of propidium iodide-labeled cells. This constitutes the first in vitro evidence to suggest that adenosine A2A receptors may be the molecular target responsible for the observed beneficial effects of caffeine consumption in the development of Alzheimer's disease.

British Journal of Pharmacology (2003) 138, 1207–1209. doi:10.1038/sj.bjp.0705185


fragment with amino acids 25–35 of β-amyloid protein



ZM 241385



Adenosine is a neuromodulator in the nervous system that refrains neuronal excitability through activation of inhibitory A1 receptors (Cunha, 2001). This has led to the exploitation of A1 receptor ligands as potential neuroprotective agents, an effort that has been undermined by the cardiovascular side effects of these agents (de Mendonça et al., 2000). Adenosine can also activate facilitatory A2A receptors, which are highly abundant in the basal ganglia and have low density elsewhere in the brain (Fredholm et al., 2002). The association of A2A receptors with dopamine D2 receptors in the basal ganglia has fostered the development of A2A receptor antagonists as anti-Parkinsonian drugs (Fredholm et al., 2002). Interestingly, it was noted that A2A receptor antagonists not only provided symptomatic relief but also arrested the progression of neurodegeneration in animal models of Parkinson's disease (Chen et al., 2001; Ikeda et al., 2002). Antagonists of A2A receptors also confer neuroprotection in other neurodegenerative situations in the basal ganglia (e.g. Reggio et al., 1999; Popoli et al., 2002) and, surprisingly, also in the extra-striatal region of the brain, namely in cortical areas (e.g. Monopoli et al., 1998; Behan & Stone, 2002). A possible in vivo correlate of this neuroprotective role of A2A receptor antagonists is the protection conferred by caffeine, an adenosine receptor antagonist, in Parkinson's disease (Schwarzschild et al., 2002). Interestingly, caffeine consumption has also recently been proposed to be a protective factor in the development of Alzheimer's disease (Maia & de Mendonça, 2002). This led us to test if caffeine and an A2A receptor antagonist could protect neurons against the neurotoxicity caused by exposure to β-amyloid protein, a purported effector of neurodegeneration occurring in Alzheimer's disease (Vickers et al., 2000).


Primary cultures of cerebellar granule cells were prepared from 8-day-old Wistar rats, as previously described (Porciúncula et al., 2001). Briefly, freshly dissected cerebella were incubated with 0.025% trypsin solution for 15 min at 37°C and disrupted mechanically in the presence of 0.08 mg ml−1 DNAse and 0.05% trypsin inhibitor. Cells were then seeded at a density of 1.5 × 105 cells cm−2 in a 96-well multiwell dish coated with 10 μg ml−1 poly-D-lysine and incubated in Eagle's basal medium supplemented with 10% fetal bovine serum, 50 μg gentamicin and 25 mM KCl. The growth of non-neuronal cells was inhibited by addition of 20 μM cytosine arabinofuranoside 18–24 h after seeding and the medium was not changed during the culture period. After 5 days of growth in a humidified atmosphere of 95% air and 5% CO2 at 37°C, cells were deprived of serum and incubated in the absence or presence of fragment with amino acids 25–35 of β-amyloid protein (Aβ25–35) (25 μM, from Sigma, Sintra, Portugal) without or with either caffeine (0.2–25 μM, from Sigma) or 8-cyclopentyltheophylline (CPT, 200 nM, from Sigma) or 4-(2-[7-amino-2-(2-furyl)(1,2,4)triazolo(2,3-a) (1,3,5)triazin-5-ylamino]ethyl)phenol (ZM 241385, 50 nM, from Tocris Cookson, Bristol, U.K.). Cellular viability was evaluated after 48 h by loading the cells for 3 min with 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 μg μl−1 propidium iodide (Calbiochem), and red-labeled cells were photographed using an inverted Nikon Diaphot Eclipse TE 300 with standard rhodamine filter set (excitation 540 nm; emission 617 nm). Pictures were then converted into black and white, and the number of white spots (cells labeled with propidium iodide) was quantified (Scion Image 4.02 software) and used as an index of cell damage. An evaluator blind to treatments performed all quantification procedures.

The values presented are mean±s.e.m. of n experiments. To test the significance of the effect of a drug versus control, a one-way variance analysis (ANOVA) was used, followed by Duncan's test. Pleqslant R: less-than-or-eq, slant0.05 was considered to represent a significant difference.


The number of propidium iodide-labeled cells (i.e. nonviable cells) in cultures from which serum was withdrawn at day 5 was low (112±15 cells per field, n=5, see Figure 1a). When the cells were incubated without serum and with 25 μM Aβ25–35 for 48 h from day 5 onwards, there was a 183±43% (n=5) increase in the number of nonviable cells (P<0.05) compared to the control situation (withdrawal of serum) (Figure 1a, b). As illustrated in Figure 1b, the coadministration of caffeine (1–25 μM) at the time of application of Aβ25–35 (25 μM) and serum withdrawal prevented, in a concentration-dependent manner, the Aβ25–35-induced increase in the number of nonviable cells. Importantly, caffeine (25 μM), which fully prevented Aβ25–35-induced neurotoxicity, did not significantly (P>0.05) change the number of nonviable cells upon withdrawal of serum without adding Aβ25–35 (n=5, Figure 1b), indicating that endogenous extracellular adenosine selectively interferes with neuronal cell death caused by the β-amyloid fragment peptide rather than by serum withdrawal.

Figure 1.

Caffeine and adenosine A2A receptor antagonists are neuroprotective against β-amyloid-induced neurotoxicity. Rat cerebellar neurons were cultured for 5 days and serum was withdrawn at day 5 for 48 h, which led to a discrete pattern of propidium iodide-labeled cells, indicative of low number of nonviable cells (a-1) and first column from the left in (b) and (c). The exposure of cells to Aβ25–35 (25 μM) in parallel with serum withdrawal increased the number of nonviable cells ((a-2), third and fourth columns from the left in (b) and (c), respectively) in comparison with the withdrawal of serum only (considered as control, 100% being the number of nonviable cells in this situation). (b) Administration of increasing concentrations of caffeine (1–25 μM) together with Aβ25–35 (25 μM) attenuated and fully prevented the Aβ25–35-induced neurotoxicity (see also (a-3)), but addition of caffeine (25 μM) failed to modify the number of nonviable cells upon serum withdrawal in the absence of Aβ25–35 (second column in (b)). The data are mean±s.e.m. of five experiments. (c) A selective A2A receptor antagonist, ZM 241385 (50 nM, second column from the right), but not an A1 receptor antagonist, CPT (200 nM, first column from the right), blocked the Aβ25–35 (25 μM)-induced neurotoxicity, but failed to modify the number of nonviable cells upon serum withdrawal in the absence of Aβ25–35 (second and third columns from the left). The data are mean±s.e.m. of three experiments. *P<0.05 versus control (first column from the left); **versus effect of 25 μM Aβ25–35 (third and fourth columns from the left in (b) and (c), respectively).

We then tried to determine which subtype of adenosine receptor was mainly responsible for the control of β-amyloid-induced neurotoxicity. As illustrated in Figure 1c, 8-cyclopentyltheophylline (CPT, 200 nM), which at this concentration is a selective A1 receptor antagonist (e.g. Kessey & Mogul, 1998), did not significantly modify the β-amyloid-induced neurotoxicity. In contrast, the selective A2A receptor antagonist, 4-(2-[7-amino-2(2-furyl) (1,2,4) triazolo (2,3-a) (1,3,5) triazin-5-ylamino)ethyl)phenol (ZM 241385) (50 nM) (see Cunha et al., 1997), almost completely prevented the Aβ25–35-induced increase in the number of nonviable cells (Figure 1c). As observed for caffeine, both CPT (200 nM) and ZM 241385 (50 nM) failed to modify the number of nonviable cells upon withdrawal of serum without adding Aβ25–35 (n=3, Figure 1c).


The present results show that caffeine prevents the β-amyloid-induced neurotoxicity in cultured cerebellar neurons of the rat and that this neuroprotective effect is likely to be because of a blockade of adenosine A2A rather than adenosine A1 receptors. This finding adds to previous reports of neuroprotective effects resulting from adenosine A2A receptor blockade in different stressful situations (Monopoli et al., 1998; Reggio et al., 1999; Chen et al., 2001; Ikeda et al., 2002; Behan & Stone, 2002; Popoli et al., 2002) suggesting that blockade of adenosine A2A receptors may be a general neuroprotective mechanism.

However, neuroprotection afforded by blockade of A2A receptors is, so far, phenomenological since the mechanism operated by A2A receptors to confer robust neuroprotection is not known. Several mechanisms have been proposed, namely control of glutamate release, of glutamate clearance by astrocytes, of inflammatory reactivity by microglia, of vascular resistance or a direct control of calcium entry or of cell cycling in neurons (discussed in Fredholm et al., 2002). The presently observed neuroprotection in cultured neurons is particularly instructive from a mechanistic point of view because it allows one to conclude that the blockade of adenosine A2A receptors directly prevents neuronal death independent of astrocytes, microglia or vascular elements. This brings into stage recent evidence indicating that A2A receptors are able to control both cell cycling in PC12 cells (Huang et al., 2001) and in cardiomyocytes (Zhao et al., 2001) and calcium entry into neurons (e.g. Gonçalves et al., 1997; Wirkner et al., 2000), thus being potentially able to control either apoptotic- and necrotic-like neuronal death. Certainly, further work needs to be carried out to test if β-amyloid-induced neurotoxicity in rat cultured cerebellar neurons mostly involves apoptotic- or necrotic-like features and to elucidate the molecular mechanisms operated by adenosine A2A receptors to control this β-amyloid-induced neuronal cell death.

The present observation that adenosine A2A receptor blockade mimics the neuroprotective effect of caffeine against β-amyloid-induced neurotoxicity strongly suggests that blockade of adenosine A2A receptors might be the target for the observed protective effects of caffeine consumption in Alzheimer's disease (Maia & de Mendonça, 2002). It is worthwhile noting that the effects of adenosine A2A receptor antagonists do not seem to desensitise over time (Xu et al., 2002) and that the extra-striatal density of adenosine A2A receptors increases in aged animals (Cunha et al., 1995), making this receptor particularly interesting as a therapeutic target. It will be interesting to see if the long-term intake of adenosine A2A receptor antagonists, now in clinical trials, will have the expected protective effect on cognitive status of the subjects enrolled in these studies.


Supported by FCT (Grants PRAXIS/SAU/44/96 and SAU/14012/1998), CNPq, CAPES and PRONEX (No. 41960904-366/96).