Purinergic signalling and immune cells

This review article provides a historical perspective on the role of purinergic signalling in the regulation of various subsets of immune cells from early discoveries to current understanding. It is now recognised that adenosine 5′-triphosphate (ATP) and other nucleotides are released from cells following stress or injury. They can act on virtually all subsets of immune cells through a spectrum of P2X ligand-gated ion channels and G protein-coupled P2Y receptors. Furthermore, ATP is rapidly degraded into adenosine by ectonucleotidases such as CD39 and CD73, and adenosine exerts additional regulatory effects through its own receptors. The resulting effect ranges from stimulation to tolerance depending on the amount and time courses of nucleotides released, and the balance between ATP and adenosine. This review identifies the various receptors involved in the different subsets of immune cells and their effects on the function of these cells.


Introduction
Although nucleotides, such as adenosine 5′-triphosphate (ATP) and uridine 5′-triphosphate (UTP), are mainly intracellular, they are released into the extracellular fluids by various mechanisms. One of them is cell damage and death: both necrotic and apoptotic cells release ATP and other nucleotides that thus constitute "danger signals" or damage associated molecular pattern [1][2][3]. In the absence of cell death, they are also released in response to various types of stress: mechanical stimulation (stretch, shear stress) [4], hypoxia or pathogen invasion [5,6]. Specific mechanisms of release include: exocytosis of secretory granules, vesicular transport [7,8] and membrane channels, such as ATP-binding cassette transporters, pannexins [9][10][11] and connexins [12]. In particular, nucleotides are released by exocytosis during platelet aggregation and synaptic transmission. For many years, cells of the immune system were not considered to be innervated, but there is increasing recognition that nerves can influence the immune system and the field of neuroimmunology is growing rapidly [13][14][15].
Once in the extracellular fluids, nucleotides are rapidly degraded by a variety of ectonucleotidases [16], such as the ENTPDases, like CD39 that degrades ATP into adenosine 5′diphosphate (ADP) and ADP into adenosine monophosphate (AMP) and CD73/5′-nucleotidase that converts AMP into adenosine. Receptors for extracellular nucleotides and their degradation products such as adenosine have been progressively characterized. Subdivision of purinergic receptors between P1 (adenosine) and P2 (ATP, ADP) was proposed by Burnstock in 1978 [17]. A further subdivision of P2 receptors between P2Y and P2X was made in 1985 [18]. It is now well established that signalling by extracellular nucleotides is mediated by these two families of receptors, the molecular structure of which has been characterized: P2Y receptors are metabotropic G protein-coupled while P2X receptors are oligomeric ion channels.
Numerous reviews on various aspects of purinergic signalling in the immune system are available (Table 1). In the history and development of knowledge about purinergic signalling, early workers focussed on adenosine, while those concerned with ATP rarely referred to adenosine. This is obviously an inadequate approach since the effects of ATP and adenosine, its breakdown product that is rapidly produced by ectonucleotidases, are intimately related. In this review, purinergic signalling in immune cells will be covered in a comprehensive and historical way, following the increase in knowledge from the early discoveries to current understanding. The review will consider the major subsets of immune cells and, for each of them, address the mechanisms of nucleotides release and adenosine generation, as well as the repertoire of functional P1 and P2 receptors that they express.

Polymorphonuclear leukocytes
Neutrophils P1 receptors Ectoenzymes that hydrolyse ATP have been observed on guinea pig polymorphonuclear leukocytes [ 1 9 -2 1 ] . I n p a r t i c u l a r, b o t h C D 3 9 a n d C D 7 3 ectonucleotidases are present on neutrophil membranes [22]. Furthermore, neutrophils express mRNA for A 1 , A 2A , A 2B and A 3 receptors [23], but the mRNA for A 2A and A 3 receptors are the most abundant [24]. Adenosine was shown to be a physiological modulator inhibiting the generation of superoxide (O 2 − ) anion by neutrophils via A 2 receptors [25][26][27][28][29]. Not surprisingly, dipyridamole, which prevents the uptake of adenosine, thereby increasing extracellular levels, inhibits O 2 − generation by neutrophils [30]. Adenosine also inhibited the degranulation induced by the chemotactic peptide N-formyl-methionyl-leucyl-phenylalanine (fMLP) [28], phagocytosis [31] and the bactericidal function of neutrophils [32]. It was proposed that the inhibitory actions of adenosine on neutrophils were due to calcium entry blockade [33][34][35]. Adenosine inhibited neutrophil respiratory bursts in association with an increase in cyclic AMP (cAMP) and reduction in [Ca 2+ ] i [36,37]. Occupancy of A 2A receptors by adenosine inhibits fMLP-induced neutrophil activation via cAMP and protein kinase A regulated events [38]. Caffeine intake results in increase in cAMP accumulation and decrease in O 2 − anion production in human neutrophils, mediated by A 2A receptors [39]. Activation of A 2A receptors also inhibited the expression and release of various cytokines and chemokines after lipopolysaccharide (LPS) stimulation of neutrophils [40]. But other studies showed that both the A 2B and the A 3 receptors Table 1 Reviews on the role of purinergic signalling in the immune system can also play a role in these inhibitory actions. Tumour necrosis factor-α (TNF-α) production by neutrophils following renal ischemia-reperfusion was increased in A 2B -deficient mice [ In contrast, there are discrepant reports concerning the action of adenosine on neutrophil chemotaxis. It was claimed in 1982 that adenosine had no effect on the chemotaxis of neutrophils, although it did enhance the inhibition of chemotaxis by 3-deaza-(±)aristeromycin [47]. However, it was reported later that adenosine promotes neutrophil chemotaxis [48], perhaps via A 1 receptors [49]. It was shown recently that the recruitment of neutrophils and other leukocytes in the lung during influenza infection is reduced in A 1 -deficient mice [50]. In contrast, LPS-induced recruitment of neutrophils in the lung was increased in A 2A -deficient mice and experiments with chimeric mice revealed that this involves a direct inhibitory effect of the A 2A receptor in myeloid cells [51]. Similar results were obtained in A 2B -deficient mice [52]. Interestingly, in A 2A -deficient mice, neutrophils were increased in the alveolar space [51], whereas they were increased in the interstitium of A 2B -deficient mice [52]. Chen et al. [24] showed that adenosine stimulates neutrophil migration and amplifies the action of chemotactic signals through A 3 receptors that are recruited to the leading edge (see Fig. 1). In A 3 -deficient mice, the recruitment of neutrophils was reduced in the lung during sepsis [53] and in the colon after induction of colitis by dextran sulphate [54]. Interestingly neutrophil chemotaxis requires excitatory signals at the front and inhibitory signals at the back of cells. This inhibitory signal at the back might be mediated by adenosine acting on A 2A receptors [55] (Fig. 1).
There is evidence that adenosine can modulate the interaction of neutrophils with pathogens. A 3 receptors aggregate in highly polarised immunomodulatory microdomains of human neutrophil membranes. They promote the formation of filipodia-like projections (cytonemes) that can extend up to 100 μm to tether pathogens. Exposure to bacteria or an A 3 agonist stimulates the formation of these projections and bacterial phagocytosis, whereas an A 3 antagonist inhibits cytoneme formation [56].
Neutrophil adherence to endothelium was enhanced via A 1 receptors and inhibited via A 2 receptors [57,58]. It is now believed that adenosine generated from ATP by CD39 and CD73 on the vascular surface functions as an anti-adhesive signal for neutrophil binding to microvascular endothelia through activation of neutrophil adenosine A 2A and A 2B receptors [59]. Activation of A 2A receptors also inhibits expression of α4/β1 integrin on human neutrophils [60]. Human neutrophils activated by fMLP increased the number of cell surface β 2 integrins on endothelial cells and induced the shedding of Lselectin. These effects were inhibited by adenosine, most likely via the A 2A receptor [61]. A 2 receptor activation inhibited neutrophil injury to coronary endothelium [62]. Adenosine also acts on endothelial receptors, thereby promoting vascular barrier function, providing a mechanism to dampen vascular leak syndrome during neutrophil-endothelial interactions [63] and regulating neutrophil chemotaxis [64]. Exposure of human endothelial cells to hypoxia/re-oxygenation caused increased neutrophil adhesion, an effect prevented by adenosine [65]. Adenosine also reduced the stimulatory effect of neutrophils Fig. 1 Proposed model of neutrophil chemotaxis. As previously reported, stimulation of chemoattractant receptors induces local release of ATP through PANX1 channels at the site that first encounters the chemoattractant. Autocrine feedback via P2Y 2 receptors amplifies the chemotactic signal and triggers cell polarization, whereby cells assume an elongated shape, and PANX1, CD39 (NTPDase1) and A 3 adenosine receptors accumulate at the leading edge. In the current study, we found that A 2A receptors are translocated from the leading edge toward the back of polarized neutrophils and that inhibitory signaling via A 2A receptordependent cAMP accumulation inhibits excitatory chemotactic signalling by blocking FPR-dependent ERK and p38 MAPK activation globally with the exception of the leading edge. ALP alkaline phosphatase, ADO adenosine, PIP3 phosphatidylinositol (3,4,5)-triphosphate. (Reproduced from [55], with permission from the American Society for Biochemistry and Molecular Biology) on tissue factor-dependent coagulant activity of endothelial cells as a result of the inhibition of neutrophil adhesion to endothelial cells mediated by A 2 receptors [66].
Adenosine might also play a role in the regulation of neutrophil number. Synergistic effects of granulocyte colony-stimulating factor and dipyridamole increased neutrophil production in mice [67]. Both effects were inhibited by adenosine deaminase (ADA). Theophylline has an immunomodulatory action on neutrophil apoptosis via A 2A receptor antagonism [68].
The expression of adenosine receptors on neutrophils can be modulated in pathological conditions and following various interventions. A 2A receptors on freshly isolated human neutrophils are upregulated after stimulation by LPS or TNF-α, and this may represent a feedback mechanism to control inflammation [69]. A 2B receptor activity in neutrophils is reduced in patients with systemic sclerosis [70]. A 4.6-fold decrease in adenosine-mediated inhibition of neutrophils from patients with septic shock was reported [71]. Hypertonic saline upregulates A 3 receptor expression on activated neutrophils and increases acute lung injury after sepsis [72]. Alterations in the functional expression of both A 2A Chen et al. [24] also showed that neutrophil ectonucleotidases hydrolyze ATP to adenosine, which, via A 3 receptors, also promoted cell migration (Fig. 1). In agreement with this concept, both P2Y 2 and A 3 receptors control the recruitment of neutrophils to the lungs in a mouse model of sepsis [53]. Neutrophil chemotaxis requires excitatory signals at the front and inhibitory signals at the back of cells that regulate cell migration. P2Y 2 receptors, as well as A 3 receptors, were shown to contribute to excitatory signals at the front, while adenosine acting on A 2A receptors contributed to the inhibitory signal at the back [55] (Fig. 1) 14 receptors was shown to be a mediator of mast cell degranulation and considered as a potential therapeutic target for allergic conditions [165]. In a recent paper, all P2Y receptor subtypes were shown to be expressed in variable levels by human LAD2 mast cells [166]. Although P2Y 4 and P2Y 11 receptors were highly expressed, they did not appear to play a major role in degranulation, whereas P2Y 14 receptors did.
Autonomic nerves as well as sensory-motor nerves innervate immune cells and release ATP as a cotransmitter in close vicinity of immune cells [167]. Indeed in accordance with the definition of the autonomic neuroeffector junction, close contact of nerve varicosities with effector cells in effect constitutes innervation, albeit of a transient nature [168,169]. Mast cells were the first claimed to be innervated [170]. Antidromic stimulation of sensory nerves increased degranulation of mast cells in the skin, and this effect was mimicked by ATP [171]. Close opposition of nerve varicosities containing small and large vesicles and mast cells in the mucosa of intestine was shown with electron microscopy [154,172] and also in cerebral blood vessels [173] (Fig. 3). Synovial mast cell activity that contributes to inflammation in joints was shown to be influenced by both unmyelinated afferent and sympathetic efferent nerves [174]. Sympathetic and trigeminal sensory nerve fibres influence rat dural mast cells and have been shown to play a role in the pathophysiology of vascular headache [175]. Functional interactions between sensory nerves and mast cells of the dura mater have been described in both normal and in inflammatory conditions [176]. Vagus nerve stimulation modulates histamine content in mast cells in the rat jejunal mucosa [177]. From a study of co-cultures of nerves and mast cells, it was concluded that ATP released from activated mast cells was an important mediator to activate nerves [178,179]. While substance P released from nerves activated mast cells, ATP released from mast cells in response to anti-IgE antibody activated superior cervical ganglia neurons. Few investigations have been carried out about the influence of nerves on non-mast cell immune cells, but evidence has been presented that nerve fibres form close relationships with other immune cells, such as eosinophils [  Hypothetical sequence of events leading to P2X 7 receptor and pannexin 1 (panx-1)-mediated inflammasome activation. Pathogen-associated molecular patterns (PAMPs) bind to Toll-like receptors (TLRs) and drive interleukin (IL)-1β gene expression and accumulation of the procytokine. Extracellular ATP binds to the P2X 7 receptor and triggers K + efflux and panx-1 activation. The functional significance of K + efflux is unknown, although it might facilitate or even precipitate inflammasome activation. Likewise, the mechanism of panx-1 activation by the P2X 7 receptor is unknown. Panx-1 in turn activates the inflammasome. Data suggest that the ion-carrying activity of panx-1 is unnecessary for inflammasome activation. The activated inflammasome then cleaves pro-IL-1β. Thus, stimulation of the inflammasome by extracellular ATP can be split into two steps: (a) recruitment and activation of panx-1 by the P2X 7 receptor and (b) activation of the inflammasome by panx-1. Colour coding: white PAMP, red TLR, green NALP3 inflammasome, orange protein-protein interaction domains, further subdivided into orange square, ASC apoptosis-associated speck-like protein containing a caspase-recruitment domain and orange octagon, pyrin domain; yellow FIIND domain, light blue caspase domain (Casp-1), dark blue biologically active IL-1β and IL-1β propiece, violet P2X 7 receptor, light green panx-1. (Reproduced from [210], with permission from Elsevier) monocytes. TNF-α production in human monocytes was inhibited by P1 receptor agonists [188]. Both A 2A and A 2B receptors were shown to be involved in the inhibition of TNF-α production [189]. Activation of A 2A receptors also inhibited IL-12 and stimulated IL-10 production by human monocytes [190,191]. These actions may contribute to suppression of Th1 responses. However, the effect of P1 receptor agonists on cytokine release from human mononuclear cells was shown to depend on the specific Toll-like receptor (TLR) subtype used for stimulation: the A 2A agonist CGS21680 inhibited TLR4-mediated TNF-α release, but potentiated TLR3-and TLR5-mediated IL-6 release [192]. Activation of A 2A receptors also inhibited LPS-induced IL-18 production, expression of adhesion molecules and production of TNF-α, in human monocytes [193,194]. Activation of A 1 receptors promoted multinucleated giant cell formation by human monocytes [195]. Adenosine analogues were shown to produce apoptosis of human mononuclear cells via A 2A and A 3 receptors [196].
P2 receptors ATP and ADP were initially shown to increase [Ca 2+ ] i in monocytes and to regulate the activity of adhesion receptors CD11b/CD18 [197]. ATP and ADP activated human promonocytic U-937 cells apparently via different P2 receptor subtypes [198]. mRNA for P2X 7 and P2Y 2 receptors was shown to be expressed by human THP-1 monocytic cells and monocytes, and the presence of these receptors was supported by pharmacological data [199][200][201]. P2X 7 receptor expression in THP-1 monocytes was positively modulated by pro-inflammatory stimuli and negatively modulated by cAMP, a classic anti-inflammatory second messenger [202]. P2X 7 receptors mediated ATP-induced IL-1β release from human and canine monocytes [203][204][205], an effect requiring priming by LPS [206]. This mechanism plays a major role in the physiological control of IL-1β secretion by monocytes. Indeed microbial components acting on different pathogensensing receptors, as well as the danger signals uric acid and C3a, induced the activation of human monocytes and their secretion of IL-1β and IL-18 through a process involving, as an initial event, the release of ATP [207][208][209]. This was followed by the autocrine stimulation of P2X 7 receptors and inflammasome activation [210] (Fig. 4). Indeed, IL-1β secretion was inhibited by apyrase as well as by P2X 7 antagonists. Additional evidence in favour of the involvement of P2X 7 was the observation that the P2X 7 receptor polymorphism Glu496Ala, which is associated with a loss of function, impaired ATP-induced IL-1β release from human monocytes [211].
ATP was initially described as a chemoattractant for monocytes [212,213]. More recently apoptotic thymocytes were found to release nucleotides leading to the recruitment of monocytes [3]. This release is mediated by pannexin 1 channels, as demonstrated by the use of pharmacological inhibitors and small interfering RNA (siRNA), and involves the activation of pannexin 1 by caspases [9]. Monocyte recruitment by apoptotic cells supernatants, demonstrated inter alia in the murine air-pouch model, was decreased in P2Y 2 −/− mice, leading to impaired clearance of the apoptotic cells. These data clearly identify ATP as a find-me signal acting through the P2Y 2 receptor that recruits monocytes in order to clear apoptotic cells. These mice were also protected against smoke-induced lung inflammation and emphysema, as a result of decreased activation of lung macrophages [267]. P2X 7 -mediated ATP-induced killing of mycobacteria by human macrophages was initially reported in 1997 [268]. This seminal observation was later confirmed in numerous studies. Mycobacterial killing involved phagosome-lysosome fusion [269] that was induced by the rise of Ca 2+ and the activation of PLD resulting from P2X 7 activation [270]. It was decreased in macrophages from P2X 7 −/− mice [271]. Infection by mycobacteria upregulated the expression of P2X 7 and its activation by ATP not only enhanced intracellular bacterial killing but also induced the apoptosis of macrophages [272] or autophagy [273]. This dual response was missing in macrophages from P2X 7 −/− mice [271]. ATP-induced bacterial killing was abrogated in macrophages from individuals homozygous for a loss of function P2X 7 polymorphism [274] and reduced by 50 % in heterozygous subjects [275]. Additional polymorphisms leading to similar consequences were described later [276]. Furthermore, the pattern of gene expression in response to ATP was different in patients with tuberculosis and controls, suggesting that a defective function of P2X 7 might lead to the development of tuberculosis [277].
Infection by parasites, such as Leishmania amazonensis [278,279] and Toxoplasma gondii [280,281], also increased the expression of P2X 7 that mediated a dual response of parasite killing and macrophage apoptosis. The P2X 7 receptor is also involved in various additional responses of macrophages. ATP released by LPS increased NOS expression and NO production in RAW 264.7 macrophages via P2X 7 receptors [282][283][284][285][286][287][288]. The P2X 7 receptor was also associated with the generation of reactive oxygen species (ROS) [289][290][291] and leukotriene B 4 [279,292]. Activation of P2X 7 receptors on macrophages induces the activation and release of tissue factor and thus favours thrombosis [293,294]. Phagocytosis of nonopsonised beads and heat-killed bacteria was increased by P2X 7 over-expression, showing that it can behave as a scavenger receptor, but this effect was inhibited by ATP [219,295]. Loss of function polymorphisms of P2X 7 and P2X 4 receptors were associated with reduced phagocytosis and were overrepresented in patients with macular degeneration [296]. P2X 7 receptors play a role in the generation of macrophage-derived giant cells, a hallmark of chronic inflammation [297]. Spontaneous cell fusion was indeed described in macrophage cultures expressing high levels of the P2X 7 receptors [298]. Furthermore, the formation of multinucleated giant cells was inhibited by P2X 7 antagonists and in macrophages from P2X 7 -deficient mice [299,300].
Despite the dominant role of P2X 7 in macrophages, evidence has accumulated to support the role of additional receptors. Multiple P2X and P2Y receptor subtypes were identified in mouse J774, spleen and peritoneal macrophages [301]. In an extensive study, mRNA for P2X 1 , P2X 4 , P2X 5 , P2X 7 , P2Y 2 , P2Y 4 , P2Y 6 , P2Y 11 , P2Y 13 and P2Y 14 receptors were all expressed by human alveolar macrophages [302]. It was suggested that other P2X receptor subtypes, in addition to P2X 7 receptors, were involved in the ATP-mediated current in human macrophages [303]. In particular it was shown that a small slowly-desensitising ATP-induced current was abolished in P2X 4 −/− mice [304]. This P2X 4 response might contribute to the P2X 7 -induced cell death that was reduced by siRNA against P2X 4 [305,306]. It has been reported that HIV binding to macrophages stimulates the release of ATP and that P2X 1 is necessary for the entry of HIV in macrophages [307]. P2Y receptors are also expressed and functional. Low concentrations of ATP were shown to activate PLC and IL-6 transcription [308]. Studies of P2Y 2 and P2Y 4 receptor knockout mice led to the conclusion that P2Y 2 receptors are the dominant P2Y receptor subtype in mouse peritoneal macrophages [309]. Nucleotides, released by apoptotic cells, through pannexin 1 [310], act as 'find-me' signals to promote P2Y 2 -dependent recruitment of phagocytic macrophages (as well as monocytes and dendritic cells (DCs)) and this recruitment is reduced in P2Y 2 -deficient mice [3]. The chemoattractant effect of C5a on macrophages was amplified by the release of ATP and the autocrine stimulation of P2Y 2 and also P2Y 12 receptors [311]. P2Y 2 receptors also mediate potentiation of prostaglandin E 2 release involved in the induction of NOS [312,313] and stimulate the production of monocyte chemoattractant protein-1 (MCP-1)/chemokine (C-C motif) ligand 2 (CCL2) [314]. Furthermore LPS potentiated nucleotide-induced inflammatory gene expression via upregulation of P2Y 2 receptors [315]. P2Y 6 receptor expression also increased following macrophage activation [309]. Indeed the amount of IL-6 and macrophage inflammatory protein-2 released in response to LPS was significantly enhanced in the presence of UDP, and this effect was lost in the macrophages of P2Y 6 knockout mice [316]. Activation of P2Y 6 receptors increased the clearance of Escherichia coli and improved survival to peritonitis through the release of MCP-1 and enhancement of macrophage chemotaxis [317]. The P2Y 11 receptor was also reported to be functional in macrophages [318]. These authors observed that ATP released from LPS-activated macrophages by vesicular exocytosis  [344], with permission from Springer) activated the P2Y 11 receptor, leading to a M1 polarisation characterized by an increased production of IL-12 [318].

P1 receptors
The first evidence of a role of adenosine and its receptors in microglia was derived from the observation of effects of propentophylline, a neuroprotective xanthine derivative that increases the extracellular concentration of adenosine by inhibiting its transport into cells [319]. Propentofylline was shown to inhibit the production of ROS by microglial cells [320,321], their uptake of amyloid precursor protein [322,323] and their proliferation and release of TNF-α [324]. Further studies showed that microglia express all subtypes of adenosine receptors. Enhanced activation of microglia associated with worsened demyelination and axonal damage was observed in A 1 receptor knockout mice subjected to experimental allergic encephalomyelitis [325]. ATP-triggered migration of microglia was inhibited in A 1 −/− as well as [326]. The A 3 receptor is also involved in microglial process extension and migration [327]. On the other hand, ATP acted as a repellent for LPS-treated microglia and induced process retraction; these actions were associated with the upregulation of A 2A receptors [328]. A 2A receptor knockout mice also displayed enhanced microglial activation in a model of experimental autoimmune encephalomyelitis (EAE) [329].
P2 receptors It was initially reported that ATP, but not ADP, induced an inward current in microglia [330], associated with an increase in cytosolic Ca 2+ [331]. Further pharmacological studies suggested that these responses were mediated by P2Y receptors [332,333]. It was later shown that the ATP effect on Ca 2+ influx was mimicked by BzATP and inhibited by oxATP, supporting the role of the P 2z or P2X 7 receptor [334]. This receptor was shown to mediate the secretion of IL-1β induced by ATP or by LPS via the release of ATP [335], and to induce microglia cell death [336] as well as microglia-mediated injury of neurons [337]. The P2X 7 receptor was also shown to be involved in microglial activation by amyloid β [338]. After nerve injury, the P2X 4 receptor was upregulated in the spinal cord and selectively expressed in microglia [339]. The tactile allodynia induced by nerve injury was suppressed by antisense oligodeoxynucleotides silencing P2X 4 receptors. Knockdown of the P2X 4 receptor by siRNA inhibited migration of microglia [340]. Following brain injury, microglia extrude processes and migrate toward sites of tissue damage. Polarisation, process extension and chemotaxis did not occur in P2Y 12 -deficient mice, while baseline motility was normal [341]. Furthermore, in living P2Y 12 -deficient mice, branch extension toward sites of cortical damage was decreased. Microglial activation leads to the downregulation of P2Y 12 receptors and the upregulation of P2Y 6 receptors [342,343]. Activation of P2Y 6 receptors by UDP stimulates phagocytosis and the uptake of microspheres.
In vivo an upregulation of P2Y 6 was observed following administration of kainic that damages neurons, leading to microglia activation. Taken together these findings show that ADP, acting through P2Y 12 , is a find-me signal for microglia, whereas UDP, acting on P2Y 6 , behaves as an eat-me signal [344] (Fig. 5).
Section summary ATP released from apoptotic cells constitutes a find-me signal that attracts monocytes/macrophages, an action mediated by the P2Y 2 receptor. It stimulates bacterial killing and macrophage apoptosis thereby contributing to decrease the bacterial and parasite burden: this action is mediated by the P2X 7 receptor. ATP also exerts a proinflammatory effect through the secretion of IL1-β, which is mediated by the P2X 7 receptor and NLRP3 inflammasome.
In contrast, adenosine exerts an inhibitory effect on monocytes/macrophages mediated by A 2A and A 2B receptors.
Multiple P1 and P2 receptors have been shown to play a role in microglia. The P2X 7 receptor is involved in IL-1β secretion and cell death. P2Y 12 , P2X 4 , A 1 and A 3 receptors stimulate process extension and migration, whereas the A 2A receptor is inhibitory. On the other hand the P2Y 6 receptor is upregulated in activated microglia and triggers microglial phagocytosis.

P1 receptors
CD39 and CD73 ectonucleotidases [345] as well as A 1 , A 2A and A 3 but not A 2B receptors [346] are expressed by human monocyte-derived DCs. In immature DCs, adenosine induced calcium transients but no increase in cAMP. This resulted in actin polymerization, chemotaxis [346] and increased expression of co-stimulatory molecules [347]. Maturation of DCs by LPS resulted in downregulation of A 1 and A 3 receptor mRNA, whereas A 2A receptors were still expressed [346]. In these mature DCs, adenosine increased cAMP and inhibited IL-12 and TNF-α production, whereas it enhanced IL-10 secretion [346,347]. These results show that adenosine can act as a chemotaxin for immature human DCs and induce their semi-maturation, characterized by a reduced capacity to induce a Th1 polarisation of CD4 + T lymphocytes [347]. Adenosine via cAMP also decreased the capacity of human DCs to prime CD8 + T cells [348].
In murine monocyte-derived DCs, adenosine also impaired maturation and inhibited the production of IL-12, leading to tolerance: this effect was mediated by the A 2B receptor instead of the A 2A receptor active in human cells [349][350][351]. IL-27 is a cytokine produced by DCs that suppresses Th1 and Th17 responses and limits inflammation in several experimental models. The suppressive action of IL-27 was mediated at least in part by the induction of CD39 in DCs and the resulting accumulation of adenosine [352]. However, the observation that adenosine could also promote the development of murine Th17 cells, via the A 2B receptor-mediated production of IL-6, added an additional complexity [353]. The A 2B receptor was upregulated in EAE and A 2B knockout mice developed less severe EAE than wild-type mice [354]. In human plasmacytoid DCs, adenosine plays a dual role by initially recruiting immature cells to sites of inflammation, an effect mediated by A 1 receptors, and by subsequently inhibiting the production of IL-6 and IFN-α, via the A 2A receptor [355].
The physiological importance of the inhibitory effect of adenosine on DCs is supported by observations on the role of ADA. Indeed the high ADA activity of DCs might help to maintain them in an active state [356]. ADA has been shown to be upregulated in DCs from non-obese diabetic (NOD) mice leading to their spontaneous activation and autoimmune T cell activation [357]. Paradoxically DCs from CD39 −/− mice exhibited impaired antigen-presenting capacity and ability to induce a Th2 response [358,359]. This resulted in decreased allergic contact hypersensitivity [358] and allergic airway inflammation [359]. This was explained not by a defect in adenosine formation but by an increased accumulation of ATP leading to the desensitization of P2Y receptors (see below).
Finally it must be mentioned that inosine has been reported to induce DCs chemotaxis independently from adenosine receptors [360]. On the other hand AMP was shown to mimic the inhibitory effects of adenosine on DCs, and these effects were maintained in CD73deficient mice and could not be explained by adenosine contamination of AMP [361]. The mechanisms of these effects remain unknown.

P2 receptors
Human DCs express mRNA for almost all known P2 receptors [345,[362][363][364] and extracellular nucleotides exert multiple effects on them ranging from chemotaxis to control of cytokine release and induction of cell death. P2Y but not P2X agonists are potent chemotactic stimuli for immature but not mature DCs [364]. Chemotaxis was associated with a rise in intracellular Ca 2+ and actin polymerization and involved the activation of G i . Allergen challenge was shown to cause acute accumulation of ATP in the airways of asthmatic subjects and mice with experimentally induced asthma that resulted in the recruitment of DCs [1]. That recruitment was mediated by the P2Y 2 receptor. Indeed, in vitro the ATP-induced migration of P2Y 2 -deficient DCs was strongly decreased as compared to DCs from wild-type mice [115]. The attraction of DCs to the lungs in a model of allergic inflammation induced by ovalbumin was also decreased in P2Y 2 −/− mice [115]. Decreased attraction of DCs to the airways might also explain the higher mortality of P2Y 2 −/− mice with lung infection by pneumonia virus of mice, as a consequence of lowered immune response and viral clearance [365]. Interestingly, the formation of ATP gradients at a site of inflammation can also inhibit transiently the migration of human DCs, via the P2Y 11 receptor, and thereby prolong the time of encounter with antigens [366]. Conversely, antagonism of the P2Y 11 receptor might improve the migration of antigen-loaded DCs to the lymph nodes. In addition to these direct effects on migration, ATP modulated the expression of chemokine receptors, with an induction of CXCR4 and a reduction of CCR5 [367], and inhibited the release of CCL2 and CCL3 chemokines [368]. Nucleotides were also shown to modulate the maturation of DCs. Schnurr et al. [369] initially reported that ATP stimulates the expression of CD83 and the secretion of IL-12 by human monocyte-derived DCs. This action was shown to be mediated by the P2Y 11 receptor and a rise in cAMP [370]. However, la Sala et al. [371] confirmed that ATP stimulates the maturation of DCs but observed an inhibitory effect on the release of IL-12 stimulated by LPS, leading to an impaired ability to initiate Th1 responses. These apparent discrepancies were resolved by the demonstration that ATP, via P2Y 11 , increased IL-12p40 but inhibited the production of IL-12p70 [372]. Furthermore ATP synergized with LPS and sCD40L to stimulate IL-10 production. This led to the conclusion that ATP, via the P2Y 11 receptor, induces a semi-maturation of DCs, characterized by an increased expression of co-stimulatory molecules and a decreased production of bioactive IL-12, leading to increased Th2 responses or tolerance. Additional studies showed that ATP via the P2Y 11 receptor produced an impressive upregulation of the expression of thrombospondin-1 and indoleamine 2,3-dioxygenase that could play a major role in tolerance [373]. A systematic study of the effect of ATP on gene expression in DCs revealed a P2Y 11 -mediated stimulatory effect on the expression of VEGF-A, that has immunosuppressive effects in addition to its angiogenic action [374], and amphiregulin, that can exert an angiogenic and tumorigenic action [375].
Other P2Y receptors were found to be expressed on monocyte-derived DCs. ATP can modulate the function of DCs directly via a cAMP increase mediated by P2Y 11 receptors and indirectly via its degradation into ADP, which acts on P2Y 1 receptors; these distinct mechanisms combine to inhibit inflammatory cytokine production, particularly IL-12, but have a differential effect on IL-10 [376]. P2Y 12 receptors are also expressed by murine DCs and their activation increased antigen endocytosis with subsequent enhancement of T cell activation [377]. UDP, but not UTP, stimulated the release of CXC-chemokine 8 from mature human DCs, via P2Y 6 receptors [378]. UTP and UDP also acted on murine DCs to mobilize intracellular Ca 2+ and to induce cytokine production [379]. Human immature monocyte-derived DCs express P2Y 14 receptors that mediate an increase in [Ca 2+ ] i in response to agonists [380]. In plasmacytoid DCs, UTP, UDP and UDP-glucose were shown to inhibit IFN-α production [381].
Like in macrophages, ATP induced in DCs the NLRP3/ASC inflammasome signalling complexes that drive proteolytic maturation and secretion of the proinflammatory cytokines IL-1β and IL-18 [382,383]. This action was mediated by the P2X 7 receptor, which is functionally expressed on DCs [384]. P2X 7 receptors were shown to be present in microvesicles shed from DCs together with IL-1β and caspase-1 and caspase-3 [385]. P2X 7 -deficient DCs fail to release IL-1β in response to LPS and ATP [386]. This might explain the resistance to allergic contact dermatitis observed in P2X 7 -deficient mice [386]. Additional P2X 7 -mediated effects of ATP on DCs include shedding of CD23 [387], release of tissue factor-bearing microparticles [388] and apoptosis [389,390].
In the intestine ATP released from commensal bacteria induced the differentiation of Th17 CD4 + cells via the activation of lamina propria CD11c + antigen-presenting cells, apparently via a P2X receptor [391]. The number of Th17 cells was increased in mice deficient in ENTPDase7, which is preferentially expressed on epithelial cells of the small intestine [392].

Section summary
ATP can exert multiple actions on DCs, mediated by distinct receptors: chemotaxis mediated mainly by the P2Y 2 receptor; semi-maturation, characterized by increased expression of costimulatory molecules and inhibition by IL-12, which is mediated by the P2Y 11 receptor and associated with a Th2 response or tolerance; induction of NLRP3/ASC inflammasome signalling complexes, mediated by the P2X 7 receptor, that leads to secretion of IL-1β and a proinflammatory effect; and enhanced antigen endocytosis mediated by the P2Y 12 receptor.
Adenosine acting on the A 2A or A 2B receptor exerts complex effects on DCs: as ATP it impairs Th1 polarisation and favours Th2 and/or tolerance, but it can also favour Th17 cell development.

Lymphocytes
T and B lymphocytes P1 receptors Adenosine was reported to cause an increase in cAMP in lymphocytes as well as in thymocytes [393][394][395][396] and to have powerful inhibitory effects on lymphocyte proliferation [397] and the immune response in humans, particularly those who have inherited deficiency of ADA [398]. The destruction of tumour cells by mouse lymphocytes was shown to be inhibited by adenosine, and this effect was potentiated by an inhibitor of ADA [399]. It was suggested that this effect of adenosine may contribute to the lack of immune response associated with ADA deficiency.
Although it was claimed that adenosine release results from the intracellular degradation of ATP to adenosine, later studies showed that extracellular adenosine is generated following the release of ATP and its extracellular breakdown [404,405]. Human B lymphocytes showed high degrading activity, while T lymphocytes were reported to be unable to degrade extracellular nucleotides [404]. However, ecto-ATPase activity was reported on cytolytic T lymphocytes [406], and E-NTPDase activity was upregulated within 15 min of T cell stimulation [407]. Furthermore, a subset of T regulatory (Treg) cells expresses CD39 and CD73 ectonucleotidases (see below). However it was suggested that CD39 is not the exclusive switch of the immune system to trigger immunosuppression, and that an alternative adenosine-generating axis is operating [408]. This axis involves the enzymes CD38 (a nicotinamide adenine dinucleotide (NAD + ) nucleosidase) and CD303a (an ecto nucleotide pyrophosphatase).
A 2A receptors were shown to be expressed on T lymphocytes [409][410][411]. A 2B receptors were also shown to be expressed on human T lymphocytes, and it was suggested that they play a role in lymphocyte deactivation by adenosine [412]. In another study, it was suggested that A 2A receptors vary in their expression on T cell functional subsets and may regulate cytokine production in activated T lymphocytes [413]. There was lower expression of A 2A receptors on B cells. A 3 receptor mRNA and protein were shown to be expressed in both resting and activated human lymphocytes and under activating conditions they are upregulated [414]. Stimulation of A 1 and A 3 receptors were reported to block the inhibitory action mediated by A 2A receptors [415]. Exposure to adenosine prior to antigenic stimulation also induced a desensitization of cAMP accumulation leading to a stronger response to antigenic stimulation [416].
Conclusive  [421]. Furthermore, the A 2A receptor contributes to the maintenance of a normal number of naive T cells by inhibiting TCR-induced activation [422].
Adenosine also inhibits T cell mobility [423], migration to lymph nodes [424] and adhesion to the endothelium [425]. The importance of the A 2A -mediated inhibitory effect of adenosine on T cells was underscored by the discovery that CD39 is selectively expressed on Treg cells (see Fig. 6) [426] that are essential for maintaining peripheral tolerance [427,428]. In human T cells, CD39 is expressed primarily by immunosuppressive Treg cells that express the Foxp3 transcription factor, and its activity is enhanced by TCR ligation [429]. CD73 is also expressed on CD4 + /CD25 + /Foxp3 Treg cells [430][431][432]. However, subsets of Treg cells expressing CD39, but not CD73, have been identified [433]. Inhibition of ADA activity further enhanced Treg-mediated immunosuppression [432].
Several studies have shown the impact of adenosinemediated inhibition of T cells in various models of disease. Defective adenosine-induced cAMP accumulation and immunosuppression were reported in T lymphocytes of patients with systemic lupus erythematosus [434,435]. A 2A receptor activation during reperfusion after ischemia protected the myocardium from infarction and this effect was dependent on an inhibition of T cell accumulation [436,437]. A 2A receptor agonists attenuated allograft rejection and alloantigen recognition by an action on T lymphocytes [438], suppressed the development of graft-versus-host disease [439,440] and attenuated experimental autoimmune myasthenia gravis [441]. CD39 and CD73 expressed on Treg cells led to a local accumulation of adenosine that protected against Helicobacter induced gastritis [442]. Treg cells suppressed contact hypersensitivity reactions by a CD39 and adenosine-dependent mechanism [425]. In other models, the action of adenosine proved to be deleterious. A 2A -deficient mice were protected from the lethal effect of sepsis, due to preserved lymphocyte function and decreased immunosuppressive IL-10 [443]. CD39 and CD73 expressed on ovarian cancer cells generate adenosine that exerts an immunosuppressive effect, which was relieved by siRNAs against CD39 and CD73 and by an A 2A antagonist [444]. In HIV infection, Treg inhibitory effects were relieved by CD39 downregulation and reproduced by an A 2A agonist [445,446]. Furthermore, a polymorphism of the CD39 gene was identified, that is associated with downregulation of CD39 and slower progression to AIDS [446].
Few studies have been performed on B lymphocytes. Accumulation of cAMP produced by adenosine in B cells stimulated by Staphyloccocus aureus suppressed IgM production [447]. On the other hand B cells coexpress CD39 and CD73 and adenosine inhibited B cell proliferation and cytokine expression [448]. Activated B cells also inhibited T cell proliferation and cytokine production [448].
P2 receptors Early reports showed that ATP protected rat lymphocytes against the loss of intracellular enzymes into the medium [449,450] and that receptors for ATP were present on lymphocytes [451]. The action of ATP on lymphocytes is complex: ATP was reported to Fig. 7 Purinergic signalling in T cell activation. Antigen recognition by T cells involves the formation of an immune synapse between a T cell and an antigen-presenting cell (APC). The immune synapse contains a large number of signalling molecules that are required for T cell activation, including T cell receptors (TCRs), MHC molecules, co-stimulatory receptors and the purinergic signalling receptors P2X 1 , P2X 4 and P2X 7 . In response to TCR and CD28 stimulation, pannexin 1, P2X 1 receptors and P2X 4 receptors translocate to the immune synapse. ATP released through pannexin 1 promotes autocrine signalling via the P2X receptors. Confinement of ATP in the immune synapse results in a powerful autocrine feedback mechanism that facilitates the signal amplification required for antigen recognition. P2 receptors expressed and ATP released by APCs may also have important roles in regulating the antigen recognition process. NFAT nuclear factor of activated T cells. (Reproduced from [474], with permission from Springer) stimulate DNA synthesis in a subpopulations of T cells [452,453], but ATP was also shown to be highly toxic to human lymphocytes and to thymocytes, causing permeabilization of the plasma membrane and cell death [454,455]. It was later shown that ATP increased cytosolic Ca 2+ in mouse thymocytes [456][457][458] and stimulated the PLC pathway in human B lymphocytes [459]. On the other hand, an ATP 4− receptor-operated sodium channel was identified on human lymphocytes [460,461]. ATP-gated channels were also identified in human lymphoblasts [462].
Important advances were made in 1994 and the following years with the identification on human lymphocytes of P 2Z , now called P2X 7 , receptors antagonised by oxATP [463] and by the isoquinoline derivative KN-62 [464]. P2X 7 receptors were also identified specifically in human B lymphocytes [465,466] and murine T lymphocytes [467]. P2X 7 receptors were implicated in the mitogenic stimulation of human T lymphocytes purified from peripheral blood [468]. ATP and the selective P2X 7 agonist BzATP caused plasma membrane depolarisation and a Ca 2+ influx in T lymphocytes. ATP or BzATP alone had no effect on lymphocyte proliferation but potentiated the action of mitogens such as anti-CD3 [468]. Transfection of lymphoid cells lacking P2X 7 receptors with P2X 7 cDNA increased their proliferation [469]. Later studies showed that TCR stimulation triggers the release of ATP through pannexin-1 hemichannels [470] and vesicular exocytosis [471], and upregulates P2X 7 expression [472]. siRNA silencing of P2X 7 inhibited T cell activation, which was also lower in C57BL/6 mice that express a poorly functional P2X 7 receptor, as compared to BALB/c mice that express fully functional P2X 7 receptors [472]. Shockwaves increased T cell proliferation through ATP release and P2X 7 activation [473]. Thus ATP released through pannexin 1 channels enhances T call activation in an autocrine manner ( Fig. 7; [474]). But it is also involved in a paracrine communication that leads to calcium waves in neighbouring lymphocytes and a reduction of T cell motility in lymph nodes that would favour T cell scanning of antigen-loaded DCs [475]. However, P2X 7 receptors also induced the shedding of L-selectin (CD62L) from T cells, which accompanies T cell activation and allows T cells to move away from lymph nodes and enter the circulation [476][477][478][479][480][481][482].
ATP induced the lysis of CD4 + thymocytes and peripheral CD4 + T cells [483] and the apoptosis of murine thymocytes [484,485]. T lymphocyte subsets express different levels of P2X 7 and high levels are associated with ATP-induced cell death [486]. P2X 7 receptor-mediated cell death was also shown to differ between different stages of murine T cell maturation [487]. Interestingly mouse Treg cells express a higher level of P2X 7 and their activation by ATP leads to their depletion [488,489]. P2X 7 −/− mice have increased levels of Treg cells [490]. The P2X 7 receptor was also involved in T cell death induced by NAD + through the ADP-ribosylating ectoenzyme, ART2. Indeed ART2-catalyzed ADPribosylation activates P2X 7 receptors [491][492][493]. In particular Treg cells express ART2 and can be depleted by intravenous injection of NAD + [494]. However, ATP (1 mM) enhanced the proliferation and immunosuppressive ability of human Treg cells, whereas it induced apoptosis of CD4 + T cells [495]. The dual action of the P2X 7 receptor on growth versus death clearly depends on the concentration of ATP, with stimulatory effects at 250 nM and inhibition at 1 mM [495]. This could be related to the existence of two states of activation of the P2X 7 receptor: cation-selective channel or large conductance non-selective pore [496]. Numerous studies have shown the importance of the lymphocyte P2X 7 receptor in various models of inflammatory diseases. In some of these models, inhibition or deficiency of P2X 7 was associated with decreased immune reactions. Mycobacterium tuberculosis infected P2X 7 −/− mice had an increased microbial burden in the lung and pulmonary infiltrates contained a higher number of Treg cells [497]. oxATP was shown to inhibit T cell-mediated autoimmunity in models of autoimmune type 1 diabetes and encephalitis in mice [498]. CD38 knockout NOD mice develop accelerated type 1 diabetes. This was corrected by coablation of P2X 7 [499]. oxATP delayed islet allograft rejection [500] and increased cardiac transplant survival in mice [501]; these effects were associated with decreased T cell activation. However in other models P2X 7 deficiency was associated with increased immune reactions, illustrating the dual role of P2X 7 receptors emphasized previously. Following oral infection with Listeria monocytogenes, P2X 7 -deficient mice showed enhanced CD8 responses in the intestinal mucosa, which can be explained by the proapoptotic effect of P2X 7 on intestinal CD8 cells [502]. Graft versus host disease was enhanced in P2X 7 −/− mice, and this is associated with T cell expansion and reduced Treg cells [503]. EAE was also exacerbated in P2X 7 −/− mice as a result of decreased apoptosis of T lymphocytes [504] and increased T cell cytokine production [505]. P2X receptors other than P2X 7 have been shown to play a role in T cell control. RT-PCR studies had shown that P2X 1 , P2X 2 and P2X 6 were expressed by murine thymocytes in addition to P2X 7 [506]. ATP released through pannexin hemichannels following TCR stimulation amplified T cell activation not only through P2X 7 receptors [472] but also via P2X 1 and P2X 4 receptors, as demonstrated by the use of siRNA [474,507]. Hypertonic saline is known to increase T cell function [508]: it acts through the release of ATP and the activation of P2X 1 , P2X 4 and P2X 7 receptors, as shown also by gene silencing [509]. Both P2X 7 and P2X 4 are also involved in the activation of unconventional βγ T cells [510,511].
Although P2X receptors and particularly P2X 7 play a major role in lymphocytes, there is some evidence for the role of P2Y receptors as well. Upregulation of P2Y 2 receptor mRNA expression was described as an immediate early gene response in activated thymocytes [512] and P2Y 2 receptors were shown to be involved in ATP-induced T cell migration [495]. The P2Y 6 receptor was shown to be expressed in activated T cells infiltrating in inflammatory bowel disease [513]. Antagonists of the P2Y 6 receptor blocked murine T cell activation [514], but these results must be interpreted with caution since T cells of P2Y 6 -deficient mice exhibited an increased activity in a model of allergic pulmonary inflammation, suggesting that the P2Y 6 receptor plays an inhibitory rather than a stimulatory role [515]. P2Y 14 receptors were shown to be functionally expressed by mouse spleen-derived T lymphocytes [516]. Adenine nucleotides inhibited CD4 + T cell activation via an increase in cAMP induced by an unidentified P2Y receptor [517]. Reproduced from [550], with permission from Springer Natural killer (NK and NKT) cells P1 receptors NK cell activity was shown to be inhibited by adenosine and A 2 receptor agonists that increase cAMP [518]. Later studies demonstrated the involvement of A 2A receptors [519]. Adenosine via the A 2A receptor inhibited IFN-γ production by NKT cells, a subset of T cells with natural killer activity [520], but increased their production of IL-4 and IL-10 [521]. Mice were protected against liver reperfusion injury and concanavalin A (ConA)-induced hepatitis by adenosine acting on the A 2A receptor on NKT cells, and this protection was abolished in A 2A −/− mice [520,522]. Sickle cell disease results in disseminated microvascular ischemia and reperfusion injury that leads to the activation of NKT cells and the upregulation of A 2A receptors [523][524][525]. Activation of A 2A receptors in NY1DD mice with sickle cell disease reduced pulmonary inflammation and injury [523]. In a phase I study, the A 2A agonist regadenoson was administered to patients with sickle cell disease and was shown to inhibit the activation of NKT cells [524]. A 2A −/− mice were protected against tumor metastasis, and this protection was associated with increased NK cell maturation and cytotoxic function [526]. On the other hand, an A 3 receptor agonist was shown to potentiate NK cell cytotoxic activity [527] and IFN-γ production [528].
P2 receptors Inhibition of human and mouse NK cell reactivity via nucleotide receptors was reported [529][530][531][532]. It was later shown that ATP inhibits cell killing by NK cells via the P2Y 11 receptor and an increase in cAMP [533]. On the other hand, NKT cells express the P2X 7 receptor, the activation of which can lead to either apoptosis or cell activation [534][535][536]. In vitro NAD induced rapid apoptosis of NKT cells that was mediated by the P2X 7 receptor, but its injection in Con A-treated mice enhanced cytokine production by NK cells and liver injury, that was decreased in P2X 7 knockout mice [534]. In CD39-deficient mice, apoptosis of NKT cells was increased leading to protection against ConA-induced liver injury [535] or hyperoxic lung injury [536].

Section summary
The release of ATP through pannexin hemichannels or vesicular exocytosis amplifies in an autocrine way the TCRmediated activation of T lymphocytes. This amplification is mediated by the P2X 7 receptor, and also by P2X 1 and P2X 4 receptors. But activation of P2X 7 can also induce T cell death. The resulting effect (activation or death) depends on the particular subset of T cells and on the concentration of ATP. Adenosine exerts inhibitory effects on T lymphocytes, which are mediated by the A 2A receptor. Treg cells overexpress the ectonucleotidases CD39 and CD73 that sequentially convert ATP into AMP and adenosine, and their immunosuppressive action is partially mediated by adenosine.

Concluding remarks
Extracellular nucleotides and adenosine exert a variety of effects on distinct subsets of immune cells via a wide spectrum of receptor subtypes (Table 2). These actions can be both stimulatory and inhibitory, and the balance between the two critically depends on the amount and time course of nucleotide release. This is consistent with the role of ATP and its degradation product adenosine as danger signals that stimulate the immune response following injury but moderate this response when it becomes excessive and deleterious.

Neutrophils and eosinophils
ATP released from neutrophils amplifies their attraction by chemotactic signals via the P2Y 2 receptor and after its degradation to adenosine via the A 3 receptor, one example of cooperation between P1 and P2 receptors. The P2Y 2 receptor is also involved in the recruitment of eosinophils in the lung during allergic inflammation. On the other hand ATP and adenosine have opposite effects on O 2 − production and other functions of neutrophils: potentiation by ATP and inhibition by adenosine.
Monocytes/macrophages and microglia ATP released from apoptotic cells constitutes a find-me signal that attracts monocytes and macrophages, an action mediated by the P2Y 2 receptor. Via the P2X 7 receptor, ATP stimulates NLRP3 inflammasome activation and IL-1β secretion by macrophages, their killing of bacteria and their apoptosis. ADP acting on the P2Y 12 receptor induces the polarisation and migration of microglia. UDP acting on the P2Y 6 receptor stimulates their phagocytic activity. ADP and UDP have, thus, a complementary action of find-me and eat-me signals, respectively, involving a cooperation between two distinct P2Y receptor subtypes. The P2X 4 , A 1 and A 3 receptors have also been shown to play a role in microglia migration, whereas the A 2A receptor is inhibitory.
Dendritic cells ATP can exert an immunostimulatory effect on DCs via P2X 7 receptor activation. But it can also activate the P2Y 11 receptor leading to a semi-maturation state characterized by the upregulation of co-stimulatory molecules and the inhibition of IL-12 production, which impairs the Th1 response and favours tolerance or a Th2 response. The balance between these opposite effects depends on the amount of ATP released and the time course of this release.
Other specific functions of DCs can be activated by nucleotides via distinct P2Y receptor subtypes: chemotaxis by the P2Y 2 receptor and antigen endocytosis by the P2Y 12 receptor. Adenosine acting on the A 2A (human) or A 2B (mouse) receptors exerts complex effects on DCs: as ATP it impairs Th1 polarisation and favours Th2 and/or tolerance, but it also favours Th17 cell development.

Lymphocytes
The release of ATP through pannexin hemichannels or vesicular exocytosis amplifies in an autocrine way the TCR-mediated activation of T lymphocytes. This amplification is mediated by the P2X 1 , P2X 4 and P2X 7 receptors. On the other hand, Treg cells over-express the ectonucleotidases CD39 and CD73 that sequentially convert ATP into AMP and adenosine, which binds to A 2A receptors on effector T cells and suppresses their function.

Neuroimmunology
Contrary to earlier beliefs, the evidence that immune cells are innervated, albeit by nerve varicosities that form occasional close appositions, is convincing. This will be important for future studies of neuroimmunology that might reveal additional roles of ATP and purinergic signalling.    Gu