Early evidence for ATP release from rabbit urinary bladder epithelial cells by hydrostatic pressure changes was presented by Ferguson et al. 6, who speculated about this being the basis of a sensory mechanism. Prolonged exposure to a desensitizing concentration of α,β-methylene ATP (α,β-meATP) significantly reduced the activity of mechanosensitive pelvic nerve afferents in an in vitro model of rat urinary bladder 7. Later, it was shown that mice lacking the P2X₃ receptor exhibited reduced inflammatory pain and marked urinary bladder hyporeflexia with reduced voiding frequency and increased voiding volume, suggesting that P2X₃ receptors are involved in mechanosensory transduction underlying both inflammatory pain and physiological voiding reflexes 8. Subsequently, using P2X₂ knockout mice and P2X₂/P2X₃ double knockout mice, a role for the P2X₂ subtype was shown to be involved in mediating the sensory effect of ATP 9. In a systematic study of purinergic mechanosensory transduction in the mouse urinary bladder, ATP was shown to be released from urothelial cells during distension, and activity initiated in pelvic sensory nerves was mimicked by ATP and α,β-meATP and attenuated by P2X₃ antagonists as well as in P2X₃ knockout mice; P2X₃ receptors were localized on suburothelial sensory nerve fibres 10. It appears that the bladder sensory DRG neurons, projecting via pelvic nerves, express predominantly P2X_2/3 heteromultimer receptors 11.

Sensory information from the urinary bladder is conveyed by both lumbar splanchnic (LSN) and sacral pelvic (PN) nerves to the spinal cord. A study comparing the mechanosensitive properties of single afferent fibres in these two pathways showed that both low and high threshold stretch-sensitive afferents were present in both pathways 12. Single unit analysis of sensory fibres in the mouse urinary bladder revealed both low- and high-threshold fibres sensitive to ATP contributing to physiological (non-nociceptive) and nociceptive mechanosensory transduction, respectively 13. It was also shown that purinergic agonists increase the excitability of afferent fibres to distension. The roles of ATP released from urothelial cells and suburothelial myofibroblasts on various bladder functions have been considered at length in several reviews 1415, and evidence presented that urothelial-released ATP alters afferent nerve excitability 16. Amiloride, a blocker of epithelial Na⁺ channels, has been shown to suppress ATP release from cultured urothelial cells by a hypotonic (mechanical) stimulus 17 or by stretch of intact bladder 18. Raising the intracellular Ca²⁺ concentration inhibits stimulation-evoked ATP release from urothelial cells 19.

ATP given intravesically stimulates the micturition reflex in awake freely moving rats, probably by stimulating suburothelial C-fibres, although other mediators are likely to be involved 20. Studies of resiniferatoxin desensitization of capsaicin-sensitive afferents on detrusor overactivity induced by intravesicle ATP in conscious rats supported the view that increased extracellular ATP has a role in mechanosensory transduction and that ATP-induced facilitation of the micturition reflex is mediated, at least partly, by nerves other than capsaicin-sensitive afferents 821. ATP has also been shown to induce a dose-dependent hypereflexia in conscious and anesthetized mice, largely via capsaicin-sensitive C-fibres; these effects were dose-dependently inhibited by pyridoxalphosphate-6-azonphenyl-2',4'-disulfonic acid (PPADS) and 2',3'-O-(2,4,6-trinitrophenyl)-ATP (TNP-ATP) 22 (Figure 2a). P2X₁ and P2X₃ receptors play a fundamental role in the micturition reflex in female urethane-anesthetized rats; P2X₃ receptor blockade by phenol red raised the pressure and volume thresholds for the reflex, while P2X₁ receptor blockade diminished motor activity associated with voiding 23. In TRPV1 receptor knock-out mice, release of ATP is significantly depressed 24 and afferent sensitivity to distension is attenuated, especially those effects mediated by low threshold fibres related to the micturition reflex, rather than the high threshold nociceptive fibres 25.

Four functionally distinct populations of bladder sensory neurons were identified with electrophysiological recordings when guinea-pig bladder was subjected to a range of mechanical stimuli (stretch, von Frey hair stroking and focal compression of receptive fields) and chemical stimuli (α,β-methylene ATP and capsaicin) 26. Four different major classes of extrinsic sensory C fibres have been identified in the guinea-pig bladder: one mediates muscle mechanoresponses and was unaffected by removal of the urothelium; another was activated by stretch and α,β-meATP and was reduced by urothelial removal; the third were stretch insensitive, but could be activated by mucosal stroking with von Frey hairs or α,β-meATP and reduced by urothelium removal; while the fourth class were stretch insensitive, but could be weakly activated by mucosal stroking, but not by α,β-meATP 26.

Despite the compelling evidence in support of purinergic mechanosensory transduction from several independent laboratories (including stimulation by α,β-meATP of 2 of the 4 sensory afferents classes described by Zagorodnyuk et al. 26), a recent paper from this group claims that urothelial release of ATP and stimulation of sensory fibres is not involved in mechanosensory transduction in the bladder, but that benzamil-sensitive stretch-activated ion channels are more likely to be involved 27. Further experiments will hopefully resolve this issue.

In rats with detrusor overactivity induced by bladder outlet obstruction, there is an increase in expression of muscarinic receptors and an increase, but to a smaller extent, of P2X₃ receptor immunostaining 28. Cyclophosphamide-induced bladder inflammation (a model for interstitial cystitis), sensitizes and enhances P2X₃ and P2X_2/3 receptor function in rat bladder sensory neurons 29. Botulinum toxin A, which has antinociceptive effects in treating interstitial cystitis, inhibits distension-mediated urothelial release of ATP in conditions of bladder inflammation 30 as well as ATP release as a cotransmitter with acetylcholine from parasympathetic nerves 31.

In summary, there is now strong evidence from several different laboratories that ATP is released from urothelial cells during distension of the bladder wall. The ATP then activates sensory nerve endings beneath the urothelium, via P2X₃ and P2X_2/3 receptors, that leads, via low threshold fibres, to modulation of the voiding reflex and via high threshold fibres to reach pain centres in the CNS.

The uroteric colic that is induced by the passage of a kidney stone causes severe pain. Immunostaining of P2X₃ receptors in sensory nerves in the subepithelial region was reported 32. Multifibre recordings of ureter afferent nerves were made using a guinea pig preparation perfused in vitro 33. Distension of the guinea-pig ureter increased spike discharge in sensory neurons, which was mimicked by ATP and reduced by ATP antagonists 33 (Figure 3a). The afferent responses consisted of both fast and slow components. The P2 receptor antagonists TNP-ATP and PPADS reduced distension-induced afferent activity (Figure 3b) and blocked the rapid and reduced the slower response to ATP, while the remaining responses were blocked by the selective A₁ receptor antagonist 8-cyclopentyl-1,3-dipropylxanthine. The ecto-ATPase inhibitor (ARL-67156) produced an increase in base-line and distension-induced sensory discharge.

Knight et al. 34 found that distending the perfused guinea-pig ureter at pressures from 20-700 cm H₂O caused a pressure-dependent release of ATP from urothelial cells, approximately 10 times the basal release levels. The ATP release was abolished by removal of the urothelium and scanning electronmicroscopy confirmed an intact urothelium after distension. ATP was not released due to activation of stretch-activated channels since gadolinium failed to affect ATP release, nor did glibenclamide, known to inhibit ATP-binding cassette proteins. However, both monensin and brefeldin A, which interfere with vesicular formation and trafficking, inhibited distension-evoked ATP release, which was Ca²⁺-dependent, indicating that ATP release from ureter urothelium might be largely mediated by vesicular exocytosis. In a recent study in our laboratory, experiments have been carried out to show that ATP is released from the human ureter upon distension (Figure 4a) and that human ureteric suburothelial sensory nerves express P2X₃ receptors 35.

The release of ATP only occurred above a threshold of 25-30 com H₂O. This is similar to the uroteric pressure threshold for pain measured by Risholm 36. In a recent review of the physiology and pharmacology of the human ureter, it was suggested that purinergic receptors might be target analgesics for the treatment of ureteral colicky pain and that an additional advantage might be facilitating spontaneous ureteral stone passage 37.

A hypothesis was proposed suggesting that purinergic mechanosensory transduction in the gut initiated both physiological reflex modulation of peristalsis via intrinsic sensory fibres and nociception via extrinsic sensory fibres 3839 (Figure 2b). Evidence in support of this hypothesis was obtained from a rat pelvic sensory nerve-colorectal preparation 40. Distension of the colorectum led to pressure-dependent increase in release of ATP from mucosal epithelial cells (Figure 4b) and also evoked pelvic nerve excitation. This excitation was mimicked by application of ATP and α,β-meATP and attenuated by the selective P2X₃ and P2X_2/3 antagonist TNP-ATP and by PPADS. The sensory discharge was potentiated by ARL-67156, an ATPase inhibitor. Single fibres analysis showed that high-threshold fibres were particularly affected by α,β-meATP. In addition to release of ATP from mucosal epithelial cells in the rat gut in response to distension (see 40), ATP has also been shown to be released from human intestinal epithelial cells in response to osmotic swelling 4142. The interactions of ATP with other mediators that activate pelvic afferent fibres in the rat colorectum, including capsaicin, 5-hydroxytryptamine (5-HT), bradykinin, prostaglandins and substance P (SP), have been described 4344. In addition, TRPV1 channels are activated and sensitised by ATP that is released during distension 4546, especially in pathological states such as colitis 474849. Carvacral, an agonist for TRPV3 channels, caused increased ATP release from colonic epithelial cells 50 and TRPV4 channels have also been shown to mediate stretch-release of ATP from urothelial cells 51. LSN and PN nerves convey different mechanosensory information from the colon to the spinal cord. Forty percent of LSN afferents responded to α,β-meATP compared with only 7% of PN afferents 52.

Purinergic mechanosensory transduction has been described in other regions of the gastrointestinal tract. For instance, α,β-meATP was shown to stimulate mechanosensitive mucosal and tension receptors in mouse stomach and oesophagus leading to activity in vagal afferent nerves 53. The sensitizing effects of P2X₃ receptor agonists on mechanosensory function are induced in oesophagitis 54. Vagal nodose (placode-derived) nociceptive fibres in guinea-pig oesophagus are exclusively C-fibres sensitive to P2X₃ receptor agonists and rarely express SP, while jugular (neural crest-derived) nociceptive fibres include both A- and C-fibres and are insensitive to P2X₃ agonists and mostly express SP 55. Adenosine has been claimed to activate a subset of nociceptive vagal sensory nerves in guinea-pig oesophagus 56. Visceral hypersensitivity may play a role in the pathogenesis of functional chest pain claimed to be of oesophageal origin. Theophylline ameliorated chest pain in 7 out of 8 patients in a clinical trial, perhaps by reducing adenosine-mediated nociception 57. Purinergic mechanosensory transduction has also been implicated in reflex control of intestinal secretion, whereby ATP released from mucosal epithelial cells acts on P2Y₁ receptors on enterochromaffin cells to release 5-HT (and ATP, which is stored and co-released with 5-HT from enterochromaffin cells 58), which leads to regulation of secretion either directly or via intrinsic reflex activity 59.

Subepithelial fibroblasts in intestinal villi are highly sensitive to mechanical stimulation and release ATP during touch or stretch and probably act as mechanosensors 60. The ATP released activates P2Y₁ receptors on surrounding cells, which leads to intercellular propagation of Ca²⁺ waves and contractions in networks of subepithelial fibroblasts and a signal to sensory nerve terminals in the villi 61. Intrinsic enteric sensory nerves express P2X₃ and P2X_2/3 receptors 6263646566. In P2X₂ or P2X₃ knock-out mice, intraluminal pressure-induced peristalsis is inhibited 6566.

ATP release and P2X₃ and P2X_2/3 receptor-mediated nociceptive sensory nerve responses were enhanced in a model of colitis consisting of administration to adult rats of an intrarectal enema of 30% trinitro benzene sulfonic acid in ethanol at a dose of 80 mg/kg body weight 67. An increase in the number of DRG neurons supplying the colorectum expressing P2X₃ receptors was also claimed and there was also a substantial increase in release of ATP with distension (Figure 4b). The excitability of visceral afferent nerves is enhanced following injury or ischemia and during inflammation, for example, in irritable bowel syndrome (IBS) 68. Under these conditions, substances are released from various sources that often act synergistically to cause sensitization of afferent nerves to mechanical or chemical stimuli. Receptors to these substances (including ATP) represent potential targets for drug treatment aimed at attenuating the inappropriate visceral sensation and subsequent reflex activities that underlie abnormal bowel function and visceral pain (see 6970). Chronic functional visceral hyperalgesia induced in a rat model for IBS, induced by colonic injection of 0.5% acetic acid, is associated with potentiation of ATP-evoked responses and an enhanced expression of P2X₃ receptors in colon-specific sensory neurons 71. In addition, activation of spinal A₁ receptors with adenosine, following breakdown of ATP, has been shown to modulate visceral hyperalgesia 72.

Non-erosive reflux disease shows the classic symptoms of gastro-oesophageal reflux, but in the absence of oesophageal mucosal injury. Visceral hypersensitivity plays an important role in the pathology of this disease 73. ATP has been found to sensitise vagal afferents to mechanical stimuli in the ferret oesophagus 54 and the protein expression of P2X₃ receptors is increased in nodose and DRG with chronic oesophageal acid exposure in a rat model 74.

In the lung, pulmonary neuroepithelial bodies (NEBs) and more recently subepithelial receptor-like endings associated with smooth muscle (SMARs) 75 have been shown to serve as sensory organs in the lung, and P2X₃and P2X_2/3 receptors are expressed on a subpopulation of vagal sensory fibres that supply NEBs and SMARs with their origin in the nodose ganglia. Quinacrine staining of NEBs indicates the presence of high concentrations of ATP in their secretory vesicles, and it has been suggested that ATP is released in response to both mechanical stimulation during high-pressure ventilation and during hypoxia 76. NEBs are oxygen sensors especially in early development, before the carotid system has matured 77. In a study of bronchopulmonary afferent nerve activity of a mouse isolated perfused nerve-lung preparation, it was found that C fibres could be subdivided into two groups: fibres that conduct action potentials at < 0.7 ms^-1 and are responsive to capsaicin, bradykinin and ATP; and fibres that conduct action potentials on an average of 0.9 ms^-1 and respond vigorously to ATP, but not to capsaicin or bradykinin 78. Both the TRPV1 receptor and P2X receptors mediate the sensory transduction of pulmonary reactive oxygen species, especially H₂O₂ and OH, by capsaicin-sensitive vagal lung afferent fibres 79.

Vagal C-fibres innervating the pulmonary system are derived from cell bodies situated in two distinct vagal sensory ganglia: the jugular (superior) ganglion neurons project fibres to the extrapulmonary airways (larynx, trachea, bronchus) and the lung parenchymal tissue, while the nodose (inferior) neurons innervate primarily structures within the lungs. Nerve terminals in the lungs from both jugular and nodose ganglia responded to capsaicin and bradykinin, but only the nodose C-fibres responded to α,β-meATP. Vagal afferent purinergic signaling may be involved in the hyperactivity associated with asthma and chronic obstructive pulmonary disease 80. Th1 and Th2 cytokines reciprocally regulate P2X₇ receptor function, suggesting a role for P2X₇ receptors in pulmonary diseases, particularly lung hypersensitivity associated with chronic inflammatory responses 81.

It has been hypothesised that tissue stress or damage in the uterine cervix during late pregnancy and parturition leads to ATP release and sensory signalling via P2X receptors 82. In support of this proposal, these authors have shown P2X₃ receptor immunoreactivity in axons in the cervix, in small and medium sized neurons in L6/S1 DRG and in lamina II of the L6/S1 spinal cord segments and increases in P2X₃ receptor expression between pregnancy day 10 and parturition (day 22/23) in the rat cervix, although not in DRG or spinal cord.

P2X₃ and P2X_2/3 receptors on sensory afferents in tooth pulp appear to mediate nociception 83848586, perhaps from ATP released by mechanical distension or inflammation of odontoblasts. Mustard oil application to the tooth pulp in anaesthetised rats produced long-lasting central sensitisation, reflected by increases in neuronal mechanoreceptive field size; TNP-ATP reversibly attenuated the mustard oil sensitisation for more than 15 minutes 87.

P2X₃ receptors are abundantly present on sensory nerve terminals in the tongue 88 and ATP and α,β-meATP have been shown to excite trigeminal lingual nerve terminals in an in vitro preparation of intra-arterially perfused rat mimicking nociceptive responses to noxious mechanical stimulation and high temperature 89. A purinergic mechanosensory transduction mechanism for the initiation of pain was considered. Taste sensations appear to be mediated both by P2Y₁ receptor-activated impulses in sensory fibres in the chorda tympani 90 and by P2X₂ and P2X₃ and, perhaps, P2X_2/3 receptors 91.

The search is on for selective P2X₃ and P2X_2/3 receptor antagonists that are orally bioavailable and do not degrade in vivo for the treatment of pain (see 9293949596). Table 1 summarises the drugs widely available. Suramin, PPADS and Reactive blue 2 have been used as non-selective antagonists at P2X₃ and P2X_2/3 receptors on nociceptive sensory nerve endings. PPADS has the advantage that it associates and dissociates approximately 100 to 10,000 times more slowly than other known antagonists 97. The trinitrophenyl-substituted nucleotide, TNP-ATP, is a very potent antagonist at both P2X₃ and P2X_2/3 receptors. A-317491 (synthesised by Abbott Laboratories) and compound RO3 (synthesised by Roche Palo Alto) are both effective P2X₃ and P2X_2/3 antagonists, the latter being orally bioavailable and stable in vivo. Antagonism of P2X₁ and P2X₃ receptors by phenol red has been reported and tetramethylpyrazine, a traditional Chinese medicine, used as an analgesic for dysmenorrhoea, was claimed to block P2X₃ receptor signalling 98.

Antisense oligonucleotides have been used to down-regulate the P2X₃ receptor, and in models of neuropathic (partial sciatic nerve ligation) and inflammatory (complete Freund's adjuvant pain, inhibition of the development of mechanical hyperalgesia as well as significant reversal of established hyperalgesia, were observed within 2 days of treatment 99100101. P2X₃ antisense oligonucleotides or antagonists appear to be less effective for treating discogenic (lumbar intervertebral disc) than cutaneous tissue pain 102. Combined antisense and RNA interference-mediated treatment for specific inhibition of the recombinant rat P2X₃ receptor appears to be promising for pain therapy 103. P2X₃ double-stranded short interfering RNA relieves chronic neuropathic pain and opens up new avenues for therapeutic pain strategies in man 104.

While P2X₃ and P2X_2/3 receptors, expressed in sensory neurons, were the predominant P2 receptor subtypes first recognised to be involved in the initiation of nociception (see 105106), it has become apparent more recently that P2Y receptors are also present 107 and that these are involved in modulation of pain transmission 108. P2Y receptors appear to potentiate pain induced by chemical or physical stimuli via capsaicin sensitive TRPV1 channels and it has been proposed that the functional interaction between P2Y₂ receptors and TRPV1 channels in nociceptors could underlie ATP-induced inflammatory pain 45. P2Y₁ receptor-mediated responses also enhance the sensitivity of TRPV1-mediated responses to capsaicin, protons and temperature in a protein kinase C-dependent manner 109. ATP-induced hyperalgesia was abolished in mice lacking TRPV1 receptors.

It has been claimed that opioids inhibit purinergic nociception in rat sensory neurons and fibres via a G protein-dependent mechanism 110. Cannabinoids act as inhibitory modulators of nociceptive responses produced by P2X_2/3 receptors 111.

There are no publications to date describing clinical evaluations of P2 receptor antagonists and related purinergic compounds for the relief of pain, although clinical trials for some compounds are in progress (see 9394). Other therapeutic approaches to pain are being considered, including the development of agents that control the expression of receptors and those that enhance ATP breakdown. Further, while it is now clear that many different cell types release ATP physiologically in response to mechanical distortion, hypoxia, and various agents, we still await clear understanding of the mechanisms that underlie ATP transport. Hopefully, when this becomes clearer, agents will be developed that will be able to inhibit ATP release, another useful way forward as a therapeutic strategy.