Role of ATP in migraine mechanisms: focus on P2X3 receptors

The pathogenesis of migraine is complex since it involves interaction between peripheral and central neuronal mechanisms as highlighted in recent reviews [1,2,3]. One unresolved issue is the origin (and mechanism) of the typical pulsatile migraine pain, which is likely based on the activation of the meningeal trigeminovascular system [4,5,6]. To generate nociceptive signalling, which is further transmitted to the spinal cord/brainstem and to the higher pain centers, the trigeminal nerve terminals in the meninges should be first depolarized to a threshold sufficient to generate spiking activity [7]. To date, a lot of depolarizing stimuli were proposed to trigger such a depolarization [8] including extracellular ATP, serotonin, endovanilloids, low extracellular pH, mechanical forces and/or changes in the ambient temperature [7, 9, 10]. In addition to produce nociceptive firing, depolarization of meningeal peptidergic C-fibers can release calcitonin gene-related peptide (CGRP), which nowadays is considered a principal contributor to migraine attacks and an important target for migraine treatment [11,12,13]. The mode of action of CGRP is multifarious because it comprises activation of immune cells, control of meningeal vessels and facilitation of trigeminal afferent activity [6, 12, 14]. In particular, one key mechanism of CGRP action is sensitization of nociceptive trigeminal ganglion neurons that become hyper-responsive to various stimuli [15]. A major component of this phenomenon could be a strong upregulation of ATP-gated P2X3 receptors of trigeminal sensory neurons [16] and is one issue discussed in the present review.

While intracellular ATP is a purine compound essential for cell energy metabolism, extracellular ATP plays the role of neuromodulator/transmitter and is a potent pain-inducing agent [17,18,19]. Extracellular ATP acts on different subtypes of widely expressed ionotropic P2X and metabotropic P2Y receptors [20]. Among them, P2X2 and P2X3 subtypes expressed in sensory neurons can mediate local depolarization of nerve terminals and initiate propagating nociceptive signalling [21,22,23]. The purinergic hypothesis of migraine originally suggested by Burnstock had considered a vascular target for the ATP action [24, 25]. Later studies have shown that ATP can directly activate meningeal afferents [26,27,28] supporting the role of an ATP neuronal mechanism in migraine headache. Furthermore, because ATP is also released by glial cells and by neurons alone [29, 30] or together with other transmitters [31], its activity may be extended to key regions of the CNS implicated in migraine.

The current focused review, drawn from our research carried out with in vitro preparations of trigeminal ganglia and meningeal tissue, discusses the trigeminal sensory mechanisms likely underlying the algesic action of ATP and the potential role of P2X3 receptors (widely expressed by such neurons [32]) in migraine pathophysiology. Thus, the present data should be considered euristically to stimulate further research in vivo on this subject and any translational value to the clinic.

Synthesis, release and degradation of ATP in migraine relevant tissues

Intracellular concentration of ATP is in the range of mM [33], while even higher levels of ATP can be found in synaptic vesicles as ATP is the co-transmitter released together with principal transmitters such as glutamate, noradrenaline, acetylcholine and GABA [31].

Apart from neuronal vesicular release, ATP can also be released from immune, vascular and glial cells or neurons through pannexin channels activated by mechanical forces or activation of specific receptors [34,35,36]. Pannexin-1 channels are functionally coupled with ATP-gated P2X7 receptors in the trigeminal ganglion [34]. Enhanced ATP release can also occur due to mechanical stimuli mediated by mechanosensitive Piezo channels expressed by neuronal and non-neuronal cells [37]. Indeed, it has recently been shown that Piezo1 channels of endothelial cells can provide flow-induced ATP release [38]. Moreover, Piezo1 channels are expressed in trigeminal neurons [39, 40] and it has been hypothesized they react to pulsatile blood flow by triggering spiking activity during a migraine attack [41]. It is tempting to speculate that direct mechanical activation of Piezo1 channels by pulsating vessels and ATP-dependent depolarization of meningeal afferents represent the basic mechanism of pulsatile migraine pain [39,40,41,42].

Extracellular ATP is very unstable and can provide only a short-lasting action as it is quickly broken down in living tissues by ectoenzymes [33] (Fig. 1). In addition to ATP breakdown to AMP by ecto-nucleoside triphosphate diphosphohydrolase-1 (NTPDase1/CD39), there are also other recently emerged extracellular enzymes (NTPDases2,3,8) including ATP-diphosphohydrolase, which can dephosphorylate ATP to the P2Y agonist ADP and the latter to the almost inactive AMP. The subsequent important step in this cascade is the degradation of AMP to physiologically active adenosine (ADO) by ecto-5'-nucleotidase/CD73 [33] (Fig. 1).

Fig. 1figure 1

ATP degradation and migraine relevant signalling via P2X3, ADP-specific P2Y1 and adenosine activated A1 and A2a receptors. Extracellular ATP is degraded by CD39 to ADP and AMP and the latter is transformed to adenosine (ADO) by CD73. In contrast to CD39, neuron-specific NTPDase2 generates ADP. ATP can activate neuronal P2X3 receptors selectively expressed in sensory neurons, whereas ADP activates metabotropic neuronal P2Y1 receptors (probably also the P2Y12/13 subtypes), which in turn depress P2X3 receptor activity, thus completing this regulatory loop. Adenosine activates either inhibitory A1 or excitatory A2a receptors. Red arrows indicate activation while the block stop line shows an inhibitory effect

Production of ADP at the first step of ATP hydrolysis can activate ADP-preferring metabotropic P2Y1, P2Y12 and P2Y13 receptors, which are expressed in trigeminal neurons and in glial cells and can modulate nociception [43]. Interestingly, ADP can also provide an inhibitory effect on pro-nociceptive P2X3 receptors in sensory neurons [44] (Fig. 1). Unlike ATP, ADP does not excite meningeal afferents [26] and likely serves as negative feedback for ATP driven trigeminal nociception triggered by ionotropic receptors. This mechanism would originate from the local expression of ADP produced by NTPDases2,3,8 rather than by NTPDase1 mediated transformation of ATP to adenosine.

Adenosine can play either an anti-nociceptive role, via inhibitory A1 receptors widely expressed in neurons, or a pain triggering effect via activation of the cAMP-coupled A2a receptor subtype (Fig. 1). The latter operates via cAMP signalling to sensitize trigeminal neurons [45,46,47]. The modulatory action of ATP breakdown metabolites (ADP or adenosine) is expected to be most efficient because of the colocalization of specific NTPDases with the key components of the meningeal trigeminal nociceptive system such as vessels and nerves fibers. Our recently proposed approach, based on the detection of extracellular phosphate after ATP hydrolysis [26], has revealed ‘hot spots’ of intense ATP release/degradation around meningeal vessels surrounded by perivascular nerves.

Thus, the localization, subtype and activity of ATP degrading enzymes, the presence of downstream extracellular ATP metabolites and expression of specific receptors should all shape the functional outcome of purinergic pain signalling in migraine. This area of research is in progress and needs further studies.

Basic properties of P2X3 receptors and their function in migraine mechanisms

The P2X3 receptor is the major ATP sensitive receptor subtype expressed in rodent trigeminal neurons (up to 80% of the whole population of trigeminal ganglion neurons in primary culture) [32]. Concerning trigeminal ganglion neurons innervating the rat dura mater, retrograde labelling revealed P2X3 or P2X2 subtype (or both) expressed in 52% neurons [48]. Extracellular ATP can operate at relatively low concentrations for activation of P2X3 receptors to which it has high affinity (EC50 ~ 1 μM) [49]. When extracellular ATP is not efficiently hydrolyzed, it can inhibit P2X3 receptor at low nanomolar concentrations through a mechanism known as the ‘high affinity desensitization’ (HAD) [50, 51]. HAD is a P2X3 specific phenomenon as it is not observed with the P2X2 receptor subtype [51]. Heteromerization of P2X3 subunits with slowly desensitizing P2X2 receptors is a common phenomenon in different types of sensory neurons [52, 53] and provides an additional transducer of nociceptive signals with lower adaptation.

The nerve fibers innervating meninges are nociceptive C- and Aδ-fibers [54,55,56] with their own repertoire of calcium, potassium and sodium channels plus nociceptive sensor proteins like P2X3 receptors [9, 26]. P2X3 receptors activated by ATP are highly expressed in nociceptive Aδ-fibers but also present in unmyelinated C-fibers [57, 58]. Indeed, in vivo topical application of ATP to rat meninges induces activation of approximately half population of C- and Aδ-fibers [28]. In the isolated rat hemiskull preparation, both ATP and the stable ATP analogue α,β-meATP (agonist of P2X1 and P2X3 receptor subtypes) induce sustained spiking activity in meningeal afferents [26, 59]. An even stronger effect of ATP is observed in mouse meningeal afferents [27]. Studies with the P2X2/3 antagonist A-317491 suggest that ATP may excite meningeal afferents via P2X3 and/or P2X2/3 receptors [26].

These data on the role of P2X2 and P2X3 receptors were obtained in in vitro conditions, when a prolonged application of exogenous ATP (or its analogues) only partially mimics the action of endogenous ATP which naturally takes place in restricted areas and is limited by the high activity of NTPDases. To overcome this experimental limitation, our modelling study [60] has simulated the action of endogenous ATP released from meningeal mast cells and has indicated that a sustained pro-nociceptive effect of ATP could be achieved via: i) multiple ATP release sites; ii) highly branched axon fibers; iii) coupling of desensitizing P2X3 receptors with slowly desensitizing P2X2 receptors; and iv) co-expression of Nav1.8 sodium channels that have fast recovery from voltage-dependent inactivation. While P2X2 receptors are less expressed in sensory neurons [32], especially in human ones [61], human P2X3 receptors recover from desensitization much faster than rodent ones [50], thus supporting a more persistent process for pain signalling.

In accordance with the International Classification of Headache Disorders (third edition), tension headache is another primary headache associated with tenderness of pericranial muscles [62]. Interestingly, injection of ATP into the trapezius muscle of a small group of healthy volunteers produces more pain compared to placebo [63]. Moreover, local injection of ATP (or a,b-meATP) into neck muscles induces strong, prolonged facilitation of nociceptive signaling in brainstem networks [64,65,66]. This effect of ATP is intensified after inhibition of ADP sensitive P2Y1 receptors [64] consistent with inhibitory control of P2X3 receptors by the ADP sensitive P2Y1 subtype [44] (Fig. 1). One possible mechanism of headache originating from neck muscles may be related to the branching of trigeminal neurons that can functionally connect intra- and extracranial areas [67, 68].

Branching of meningeal afferents could also contribute to enhanced antidromic sensory spiking by supplying signalling from axon collaterals or the trigeminal ganglion itself [69,70,71]. Antidromic spiking is supposed to initiate local CGRP release, vasodilation, and degranulation of mast cells, all events which are leading to sterile meningeal neuroinflammation [69]. Our recent study has provided direct evidence that spiking activity can actually be propagated from central trigeminal fibers to the peripheral terminals in the meninges [

留言 (0)

沒有登入
gif