The A₂ receptors have been defined on the basis of their ability to stimulate adenylyl cyclase activity. Thus, they probably interact with G₅ proteins, although stimulation of phosphoinositide-dependent phospholipase C activity has been reported as well.⁶ Differently from the A₁ subtype, which can be detected in all brain areas (although to discrete levels of expression), the A_2A receptor subtype is particularly abundant and almost exclusively located to dopamine-rich brain areas. such as corpus striatum and caudatus putamen. This receptor has been involved in regulation of motor functions.² More recently, the presence of A_2A-like receptors in brain areas other than basal ganglia has been reported.⁷ These receptors have been proposed to play an entirely different role and to be responsible for stimulation of neurotransmitter release. The A_2B receptor has low affinity for adenosine and, differently from the A₁ and A_2A subtypes, which are activated by nM nucleoside concentrations, but similarly to the A₃ subtype (see below), requires micromolar concentrations of adenosine to be activated. The A_2B receptor can be found in low density in almost all cell types, including brain astrocytes. Recent data indicate that this receptor participates in the regulation of the interleukin 6 gene in astrocytoma cells,⁸ therefore suggesting a role for this receptor subtype in neuropathological conditions characterized by increased brain levels of this cytokine (e.g., Alzheimer’s disease and brain ischemia).

We now have substantial information on the mechanisms at the basis of adenosine neuroprotective actions. Following trauma and ischemia, brain concentrations of adenosine are strongly and transiently increased from resting values of 50–300 nM to 10–50 μM. Notably, during ischemia, there is a dramatic increase of extracellular adenosine in the affected tissue, mainly as a result of intracellular catabolism of ATP and subsequent transport out of the cells by several specific carrier-mediated membrane transporters.¹ There is evidence that during ischemia/hypoxia and trauma, not only decreased energy supply but also enhanced energy demand by the supraphysiological activation of excitatory glutamatergic neurotransmission leads to increased adenosine release, as a result of massively increased release of neurotransmitters from brain terminals.¹⁸ Under these conditions, adenosine contributes to limiting the deleterious consequences of sustained activation of glutamate receptors by both reducing further release of neurotoxic mediators via presynaptic A₁ receptors, and by counteracting the postsynaptic actions of excitatory transmitters, including the excessive entry of Ca²⁺ ions, which has been associated to the development of programmed cell death.

The role of the adenosine A₃ receptor in ischemia has been recently investigated by von Lubitz and co-workers. These authors showed that the in vivo chronic administration of IB-MECA dramatically improved the histopathological and neurological outcome and preserved short-term memory following cerebral ischemia in gerbils.¹⁹ Chronic IB-MECA was also protective in chemically induced (N-methyl-D-aspartate or pentamethylenetetrazole) seizures.²⁰ Significant improvement of seizure latency, neurological impairment and survival was observed. At present, it is unknown whether the protective effects of chronically administered IB-MECA are due to effects on blood flow, neurons and/or other brain cells (see below).

In recent years, our laboratory has been particularly interested at characterizing the functional roles of adenosine receptors on astrocytes, a cell type which is now recognized to play important roles in both central nervous system development and in brain repair following trauma and ischemia (for review, see Refs. ^{21, 22}). Not only do astroglial cells provide a metabolic support to neurons, but they also exert a number of highly specific functions. In developing brain, during specification of cortical areas, radial glia offer substratum and guide to migrating neurons and at the end of neurogenesis transforms into type 1 astrocytes. In adult brain, astrocytes express neurotransmitter receptors and uptake systems and directly participate in neurotransmission. Moreover, astroglial cells arc known to respond to various types of injury by rapid and vigorous astrogliosis, a reaction characterized by both increased astroglial cell proliferation and astrocytic hypertrophy, as shown by “stellation” and increased expression of glial fibrillary acidic protein (GFAP), the astrocyte-specific intermediate filament protein. Although there is still debate about whether reactive astrogliosis is beneficial or detrimental to neuronal repair mechanisms, it is known that activated astrocytes are needed for axonal growth and guidance.²³

On these bases and in keeping with the well documented protective actions of adenosine analogues, we undertook a study aimed at investigating the effects of the adenosine A₁/A₃ nonselective agonist 2-chloroadenosine (2-CA) on rat brain primary astrocytes. Choice of this agonist was simply due to the fact that it is relatively hydrolysis-resistant, and therefore seemed to us particularly suited for long-term “trophic” studies. Surprisingly, exposure of cultures to 2-CA resulted, 48–72 hr later, in a marked reduction of astrocytic cell number, as shown by a dramatic decrease of GFAP-positive cells.²⁴ 2-CA-induced decrease of astrocytic cell number was not counteracted by either xanthine derivatives or blockers of the adenosine transport system, which suggested the involvement of a xanthine-insensitive extracellular receptor. In an attempt to elucidate the molecular mechanisms responsible for 2-CA-induced reduction of astrocytes, we preincubated cultures with bromodeoxyuridine (BrdU) and then double-labeled cells with both an anti-GFAP antibody (to detect astrocytes) and an anti-BrdU antibody (to quantify the percentage of astrocytes undergoing DNA synthesis). We were surprised to find out that, despite the marked reduction of cell number, 2-CA significantly increased the percentage of cells showing anti-BrdU immunoreactivity in nuclei.²⁴ Other examples had been previously reported in the literature where cell death was shown to be accompanied by a concomitant “paradoxical” increase of DNA synthesis. This phenomenon has been well characterized for secretory epithelial prostate cells that undergo extensive apoptosis following castration. Colombel & co-workers demonstrated that prostatic cells undergoing apoptotic death incorporate BrdU into nuclear DNA prior to DNA fragmentation.²⁵ Based on these data, they concluded that quiescent epithelial cells undergo apoptosis as a result of two sequential events initiated by testosterone depletion. The first event is an active reentry of these cells into the cell cycle. The second event is the apoptotic destruction resulting from the inability of differentiated cells to successfully complete this cycle (“abortive mitosis”).²⁵ Moreover, a number of data has been accumulating in recent years suggesting that two strikingly different cellular programs as cell proliferation and cell death by apoptosis indeed share initial identical patterns of protooncogene expression,^26–29 so that the same primary stimulus activates a common pathway that can lead to either event; the timing of a secondary growth stimulus determines the final outcome.³⁰

On this background, we decided to verify whether astroglial cell death induced by 2-CA was apoptosis. Flow cytometric analysis of propidium iodide-stained nuclei showed the appearance of an hypodiploid DNA peak in cultures exposed to 2-CA that was not present in control cultures, suggesting induction of apoptosis, which was also confirmed by light and transmission electron microscopy analysis of cells.¹¹ An example of 2-CA-induced apoptosis is shown in Figure 1. This study represented the first demonstration in favor of a novel action for adenosine (induction of cell death by apoptosis) in the central nervous system.

More recently, we extended this study to the investigation of the receptor subtype responsible for adenosine-induced apoptosis. A possible role for the A₃ receptor was indirectly suggested by the demonstration that 2-CA effects were not reversed by xanthine antagonists.²⁴ We therefore tested the effects of the selective A₃ receptor agonist IB-MECA and its more recent 2-chloro derivative, 2-C1-IB-MECA.³² Exposure of rat astrocytes to high (μM) concentrations of either agonist resulted in development of apoptosis, as shown by both morphological and flow cytometric criteria.³³ On both rat astrocytes and human astrocytoma cells (ADF cells),³⁴ at concentrations 2–3 orders of magnitude lower (10–100 nM), these same agonists induced a marked reorganization of the cytoskeleton, with appearance of stress fibers and numerous cell protrusions. These morphological changes were accompanied by a significant reduction of the number of spontaneously detached apoptotic cells in the culture medium.³³

In lymphoid cells, the apoptotic effect of 2-chloro-2′-deoxyadenosine (2-CdA, cladribine, which is utilized as an anticancer agent) has been shown to require intracellular phosphorylation to 2-CdATP (cladribine-5′-triphosphate) which directly competes with dATP for DNA polymerase binding, being incorporated during DNA synthesis and repair to arrest chain extension.³⁶ DNA strand breakage is the immediate consequence of 2-CdA exposure and is more likely the primary lesion for apoptosis in these cells.^37–39 Relief from deoxyadenylate stress imbalances was obtained by stably transfecting cells with the antiapoptotic molecule bcl-2.⁴⁰

In chick embryonic sympathetic neurons, adenosine (1–100 μM) inhibited neurite outgrowth and killed about 80% of cells; also in this case, adenosine toxicity required the nucleoside to enter the cells and to be phosphorylated, as suggested by prevention of toxicity by both inhibitors of adenosine membrane transporter and by inhibitors of adenosine kinase.⁴¹

In human thymocytes, adenosine analogues were demonstrated to induce cell death via at least two independent pathways, one requiring the activation of an “atypical” extracellular receptor subtype linked to stimulation of adenylyl cyclase and Ca²⁺ increases,⁴² the other one being independent of extracellular receptors for adenosine.⁴³ For example, the effects of 2-CdA were prevented by addition of the nucleoside transport inhibitor dipyridamole, suggesting that 2-CdA has to be taken up by cells in order to trigger cell death.⁴³ On the contrary, but similar to astrocytes,³¹ the lethal effects of 2-CA were unaffected by dipyridamole, suggesting an extracellular site of action for the nucleoside. This conclusion is in agreement with the recent demonstration that both IB-MECA and CI-IB-MECA can induce apoptosis of HL-60 human promyelocytic leukemia cells.⁴⁴ This effect, similar to 2-CA-induced astroglial cell death, was not counteracted by xanthine blockers, further confirming an atypical extracellular site of action for these adenosine derivatives.⁴⁴

Adenosine-induced apoptosis of thymocytes may play a critical role in the intrathymic cell selection process which functions to prevent autoimmunity and to remove thymocytes expressing functionally inactive T cell receptors. It may be hypothesized that a disregulation of adenosinergic mechanisms in this crucial phase of T cell development may lead to serious immune system dysfunctions, as also documented by the adenosine-deaminase immunodeficiency syndrome.^45,46 Moreover, confirmation of an active role of the adenosine A₃ receptor in regulating apoptosis in lymphoid cells⁴⁴ may lead to the development of novel anticancer agents of potential value in the treatment of leukemias and other tumors.

Adenosine-regulated apoptosis of nervous system cells may have important pathophysiological implications as well. Purine-induced cell death could play a role in naturally occurring apoptosis during brain development.⁴⁷ Additionally, this mechanism could play a role in the remodeling of the brain circuitries following trauma and ischemia, by favoring the elimination of irreversibly damaged cells in order to spare energy and space for recovering ones.²² Moreover, in contrast to necrosis, where cellular contents (including neurotoxic mediators) are massively released in the extracellular space, cell death by apoptosis occurs with preservation of cell membrane integrity; cellular contents are safely sealed until phagocytosis intervenes and death occurs with no inflammation.⁴⁸ It might be speculated that adenosine-induced apoptosis in infarcted brain may function as a signal to limit the spreading of cell death following ischemic stroke.