Abstract Adenosine and ATP, via P1 and P2 receptors respectively, can modulate pain transmission under physiological, inﬂammatory, and neuropathic pain conditions. Such inﬂuences reﬂect peripheral and central actions and effects on neurons as well as other cell types. In general, adenosine A1 receptors produce inhibitory effects on pain in a number of preclinical models and are a focus of attention. In humans, i.v. infusions of adenosine reduce some aspects of neuropathic pain and can reduce postoperative pain. For P2X receptors, there is a signiﬁcant body of information indicating that inhibition of P2X3 receptors may be useful for relieving inﬂammatory and neuropathic pain. More recently, data have begun to emerge implicating P2X4 , P2X7 and P2Y receptors in aspects of pain transmission. Both P1 and P2 receptors may represent novel targets for pain relief.
Keywords Adenosine · ATP · Intravenous adenosine infusion · P2X receptors · P2Y receptors
Both adenosine and ATP are able to inﬂuence nociceptive transmission by their functions as extracellular signaling molecules and actions on cell sur- face receptors. Adenosine acts at several P1 receptors (A1, A2A, A2B, A3), all of which are coupled to G proteins (Fredholm et al. 2001). Activation of adenosine receptors can inﬂuence nociception at peripheral, spinal, and supraspinal sites, and affect nociceptive, inﬂammatory, and neuropathic pain states (Sawynok 1998; Dickenson et al. 2000; Sawynok and Liu 2003). The role of adenosine A1 receptors (A1Rs) in suppressing pain transmission is prominent, and adenosine receptors as a potential therapeutic target for the treatment of pain, either by directly acting agents (e.g., A1R agonists) or indirectly act- ing agents that increase endogenous adenosine availability (e.g., adenosine kinase inhibitors), have received considerable recent attention (Kowaluk and Jarvis 2000; McGaraughty et al. 2005). ATP acts on two families of recep- tors, the P2X ligand-gated ion channels and the P2Y metabotropic G protein- coupled receptor family (North 2002). P2X3 receptors are selectively localized on sensory nerves, and have received much attention in the context of pain signaling and the potential therapeutic development of antagonists for such receptors (Jarvis and Kowaluk 2001; Jacobson et al. 2002; Kennedy et al. 2003). More recently, it has been recognized that P2X4 receptors in the spinal cord may contribute to neuropathic pain (Tsuda et al. 2003), while P2X7 receptors may inﬂuence inﬂammatory and neuropathic pain (Chessell et al. 2005). Fur- thermore, P2Y receptors may play a role in peripheral pain signaling. The purpose of the present chapter is to focus on recent developments that ad- vance an understanding of the role of purines in nociception (see also Liu and Salter 2005).
P1 Receptors and Pain
Peripheral Influences on Nociception
Adenosine can produce different effects on peripheral pain signaling depend- ing on the nature of the receptor involved, the localization of the receptor, and the tissue conditions (i.e., normal tissue, inﬂammation, or following nerve injury). Effects on adenosine A1Rs have received the most attention, as such actions lead to suppression of pain, and there is the potential for therapeutic agents to engage such mechanisms in producing analgesia.
Adenosine A1 Receptors
The presence of A1Rs in dorsal root ganglia (Schulte et al. 2003) and trigemi- nal ganglion neurons (Carruthers et al. 2001) has now been visualized directly using immunohistochemistry. When sensory neurons are examined in vitro, adenosine A1Rs lead to reduced Ca2+ entry (Haas and Selbach 2000), decreased
cyclic AMP generation, and decreased release of calcitonin gene-related pep- tide (CGRP) (Carruthers et al. 2001). In functional studies, the local peripheral inhibitory actions of A1R agonists administered locally to the rat hindpaw
are most clearly observed as reduced mechanical hyperalgesia to inﬂamma- tory agents (Taiwo and Levine 1990; Aley et al. 1995) and reduced thermal hyperalgesia following nerve injury (Liu and Sawynok 2000). These peripheral
antinociceptive actions have been attributed to decreased cyclic AMP produc- tion in sensory nerve endings (Taiwo and Levine 1990; Carruthers et al. 2001). Repeated administration of an A1R agonist produces tolerance and cross- tolerance and cross-dependence with other agents, and it was proposed that
peripheral A1Rs exist as a complex with μ-opioid and α2-adrenergic receptors
(Aley and Levine 1997).
Endogenous adenosine can be released from sensory afferent nerves fol- lowing their activation. Thus, capsaicin (which activates TRPV1 receptors selectively expressed on C-ﬁbers; Liu et al. 2001), formalin (which leads to neu- rogenic and tissue inﬂammation; Liu et al. 2000) and glutamate (Liu et al. 2002) all increase peripheral extracellular levels of tissue adenosine as determined using peripheral microdialysis; in each case, this is inhibited by pretreatment with capsaicin. Inhibitory activity of this adenosine is revealed by the ability of a selective A1R antagonist to enhance formalin-evoked behaviors (Sawynok et al. 1998; Aumeerally et al. 2004). Peripheral administration of inhibitors of adenosine kinase also increase tissue levels of adenosine (Liu et al. 2000) and inhibit formalin-evoked responses (Sawynok et al. 1998) but not hypersen- sitivity to carrageenan inﬂammation (McGaraughty et al. 2001). Collectively, such observations indicate that inhibitory adenosine A1Rs on sensory afferents can be activated under certain conditions (i.e., mild intensities of stimulation, co-presence of certain mediators, nerve injury) by directly acting agonists, as well as by agents that increase the local tissue availability of extracellular adenosine, to produce a suppression of pain.
Adenosine A2A, A2B , and A3 Receptors
Adenosine A2A receptors are also present in the dorsal root ganglia of sensory neurons (Kaelin-Lang et al. 1998). In functional studies, local peripheral ad- ministration of A2AR agonists to the rat hindpaw leads to increased mechanical hyperalgesia (Taiwo and Levine 1990) and increased ﬂinching in response to
formalin (Doak and Sawynok 1995). Hyperalgesia is mediated by increases in cyclic AMP in the sensory nerve, which results in activation of protein kinase A, phosphorylation of Na+ channels, increased currents, and sensory afferent activation (Gold et al. 1996). Adenosine A2B and A3 receptors are present on mast cells and can lead to enhanced pain signaling by increased release of mast cell mediators such as histamine and 5-hydroxytryptamine
(Sawynok 1998). Local administration of agonists for A2ARs, A2BRs, and A3Rs also produces edema which involves mast cell degranulation (Sawynok et al.
2000; Esquisatto et al. 2001), while an A2R agonist enhances plasma extrava- sation when administered into the knee joint (Green et al. 1991). These cuta- neous and joint effects are generally regarded as reﬂecting pro-inﬂammatory actions.
Central Influences on Nociception
Adenosine A1 Receptors
The spinal administration of A1R agonists to rodents produces antinocicep- tion in models of nociceptive, inﬂammatory, and neuropathic pain, and such actions generally feature prominently in accounting for behavioral effects fol- lowing systemic administration (Sawynok 1998; Dickenson et al. 2000). The involvement of A1Rs is conﬁrmed by demonstrating antagonism of such ac- tions by selective A1R antagonists (e.g., Lee and Yaksh 1996; Poon and Sawynok
1998; Gomes et al. 1999). Recently, immunohistochemical studies have shown that A1Rs are concentrated in laminae I and II of the dorsal horn of the spinal cord (Ackley et al. 2003; Schulte et al. 2003) and are present on intrinsic dorsal horn neurons (Schulte et al. 2003); this has also been demonstrated using elec- trophysiological approaches in vitro (Hugel and Schlichter 2003). Such studies conﬁrm earlier reports which used autoradiographic approaches combined with selective lesions. While rhizotomy did not reveal a prominent presynaptic population of A1Rs on sensory nerve terminals (Geiger et al. 1984), there is some evidence for presynaptic receptors, as dorsal root ligation resulted in some accumulation of A1R immunoreactivity on the side close to the dorsal root ganglion (Schulte et al. 2003).
The main mechanisms implicated in spinal A1R-mediated antinociception include: (1) increased K+ conductance and hyperpolarization of dorsal horn intrinsic neurons (Lao et al. 2001; Patel et al. 2001; Salter and Sollevi 2001), (2) inhibition of peptide release (substance P, CGRP) (Sjölund et al. 1997; Car-
ruthers et al. 2001; Mauborgne et al. 2002), and (3) inhibition of glutamate release (Patel et al. 2001; Ackley et al. 2003; but see Yamamoto et al. 2003). A1R mechanisms also lead to a decreased release of γ-aminobutyric acid (GABA) from interneurons in the dorsal horn (Hugel and Schlichter 2003), but the
net effect of such an action is likely to be stimulatory, and its contribution to suppression of pain is not readily apparent.
Spinal adenosine systems appear to exhibit a unique efﬁcacy in models of hypersensitivity, such as neuropathic pain; this proﬁle of activity is of considerable interest, given that neuropathic pain can be difﬁcult to con- trol pharmacologically. An initial study emphasized the potency of intrathecal (i.t.) adenosine analogs in reducing manifestations of hypersensitivity com- pared to nociceptive tests (Sosnowski and Yaksh 1989). Subsequent studies conﬁrmed prominent activity in models of nerve injury with such analogs, and the potential for adenosine systems to represent a target in neuropathic pain conditions has been emphasized (Dickenson et al. 2000). Spinal admin- istration of adenosine itself, while lacking activity in normal animals, leads to relief of hypersensitivity (mechanical allodynia) following nerve injury (Gomes et al. 1999); such efﬁcacy was surprisingly long-lasting (i.e., to 24 h; Lavand’homme and Eisenach 1999). Altered pharmacokinetics did not appear to account for the duration of activity, and there was no change in the num- ber of A1Rs or alteration in G protein coupling in the spinal cord following nerve injury (Bantel et al. 2002a, b). However, i.t. adenosine (via an A1R) enhances release of noradrenaline in vivo following nerve injury (but not in normal conditions; Bantel et al. 2003), and antiallodynic actions of adeno- sine are dependent on spinal adrenergic mechanisms following nerve injury
(Gomes et al. 1999). Given the activity of α2-adrenergic receptor agonists in nerve injury states, this may contribute to the efﬁcacy of adenosine under
such conditions. The mechanism underlying this essentially excitatory action mediated by an adenosine A1R is not clear in view of prominent inhibitory actions of A1Rs. However, this action emphasizes the potential complexity of
A1R actions following nerve injury, where a number of spinal changes in no- ciceptive signaling occur (e.g., central sensitization, disinhibition, phenotype switch).
Spinal adenosine A1Rs can also be activated by agents which lead to in-
creased endogenous levels of adenosine (i.e., inhibition of adenosine meta- bolism). Thus, i.t. delivery of prototype nucleoside inhibitors of adenosine ki- nase produces intrinsic antinociception or enhanced antinociception (induced by adenosine itself, or morphine which releases adenosine; Keil and DeLander
1994; Poon and Sawynok 1998). A series of novel non-nucleoside inhibitors of adenosine kinase (A-134974; A-286501, ABT-702) produces antinociception in
inﬂammatory and neuropathic pain states following systemic administration
(Kowaluk et al. 2000; McGaraughty et al. 2001, 2005; Zhu et al. 2001; Jarvis et al.
2002c). The antinociceptive actions of adenosine kinase inhibitors occur pri- marily at spinal sites, as intracerebroventricular (supraspinal) and intraplantar (peripheral) actions were weaker compared to i.t. delivery with neuropathic pain (Zhu et al. 2001). In an inﬂammatory model, peripheral sites were not implicated in analgesic actions (McGaraughty et al. 2001). Inhibition of adeno- sine metabolism results in enhanced extracellular tissue levels of adenosine
in the spinal cord (Golembiowska et al. 1996), and this subsequently activates inhibitory spinal A1Rs (McGaraughty et al. 2005).
Adenosine A2A, A2B , and A3 Receptors
Within the central nervous system, A2ARs exhibit a high expression in basal ganglia and the olfactory bulb but low levels in other brain regions, while A2BRs and A3Rs occur at very low levels (Fredholm et al. 2001). There are some reports that spinal (Lee and Yaksh 1996; Poon and Sawynok 1998) and supraspinal (Regaya et al. 2004) administration of CGS21680, an A2AR agonist, produces antinociception, but it is weakly active compared to A1R agonists and, in the absence of effects of selective A2AR antagonists, it is not clear if this receptor is indeed involved in such actions. Furthermore, spinal electrophysiological actions of CGS21680 are complex (Lao et al. 2001; Patel et al. 2001), and the receptors mediating its effects are far from clear.
Other approaches reveal additional inﬂuences on nociception. Thus, caf- feine is a nonselective A1R, A2AR, and A2BR antagonist that exhibits analgesic
and adjuvant analgesic effects (Sawynok 1998); as inhibition of A1Rs cannot account for such activity, the analgesic proﬁles of A2AR and A2BR antagonists have received attention. Antinociception (hot plate test) has been reported
following spinal (but not systemic) administration of the A2AR antagonist SCH-58261 (Bastia et al. 2002). Furthermore, a series of A2BR antagonists was shown to produce antinociception (hot plate test) when administered systemi-
cally; as a structure that did not enter the central nervous system also exhibited activity, this action was attributed to a peripheral site of action (Abo-Salem et al. 2004). That same study observed that an A3R antagonist (PBS-10) pro- duced hyperalgesia (hot plate) following systemic administration, but that
study did not further explore this action in terms of central versus peripheral sites.