The TRPA1 channel in migraine mechanism and treatment


  • S Benemei,

    1. Clinical Pharmacology Unit, Department of Health Sciences, University of Florence, Florence, Italy
    2. Headache Centre, Department of Health Sciences, University of Florence, Florence, Italy
    Search for more papers by this author
  • C Fusi,

    1. Clinical Pharmacology Unit, Department of Health Sciences, University of Florence, Florence, Italy
    Search for more papers by this author
  • Gabriela Trevisan,

    1. Clinical Pharmacology Unit, Department of Health Sciences, University of Florence, Florence, Italy
    Search for more papers by this author
  • Pierangelo Geppetti

    Corresponding author
    1. Headache Centre, Department of Health Sciences, University of Florence, Florence, Italy
    • Correspondence

      Pierangelo Geppetti, Department of Health Sciences, University of Florence, Viale Pieraccini 6, 50139-Florence, Italy. E-mail:

    Search for more papers by this author


Migraine remains an elusive and poorly understood disease. The uncertainty is reflected by the currently unsatisfactory acute and prophylactic treatments for this disease. Genetic and pharmacological information points to the involvement of some transient receptor potential (TRP) channels in pain mechanisms. In particular, the TRP vanilloid 1 (TRPV1) and TRP ankyrin 1 (TRPA1) channels seem to play a major role in different models of pain diseases. Recent findings have underscored the possibility that TRP channels expressed in the nerve terminals of peptidergic nociceptors contribute to the migraine mechanism. Among this channel subset, TRPA1, a sensor of oxidative, nitrative and electrophilic stress, is activated by an unprecedented series of irritant and pain-provoking exogenous and endogenous agents, which release the pro-migraine peptide, calcitonin gene-related peptide, through this neuronal pathway. Some of the recently identified TRPA1 activators have long been known as migraine triggers. Furthermore, specific analgesic and antimigraine medicines have been shown to inhibit or desensitize TRPA1 channels. Thus, TRPA1 is emerging as a major contributing pathway in migraine and as a novel target for the development of drugs for pain and migraine treatment.

Linked Articles

This article is part of a themed section on the pharmacology of TRP channels. To view the other articles in this section visit


4α-phorbol 12,13-didecanoate


calcitonin gene-related peptide


chemotherapeutic-induced peripheral neuropathy


dorsal root ganglia


medication overuse headache


N-acetyl-p-benzoquinone imine


neurokinin A


phosphatidylinositol (4,5) bisphosphate


substance P


trigeminocervical complex


trigeminal ganglia


transient receptor potential


TRP ankyrin 1


TRP canonical


TRP melastatin


TRP mucolipin


TRP polycystin


TRP vanilloid


vagal ganglia

Calcitonin gene-related peptide (CGRP) and migraine


Migraine is a common and disabling neurovascular disorder, with heritability estimates as high as 50% and with a likely polygenic multifactorial inheritance (Pietrobon and Moskowitz, 2013). Migraine is characterized by attacks of often throbbing and frequently unilateral severe headache, which are usually associated with nausea, vomiting, and/or sensitivity to light (photophobia), sound (phonophobia), or odours (osmophobia), and aggravated by movement. If untreated, attacks typically last 4–72 h. In about 30% of patients, migraine attacks are preceded or accompanied by transient focal neurologic symptoms, which are usually visual, but that could also consist in paresthesias or language disturbances, commonly known as ‘aura’. The last update of the World Health Organization, Global Burden of Disease, states that migraine alone is responsible for almost 3% of disability attributable to a specific disease worldwide. In particular, migraine ranks first among neurological disorders, seventh among non-communicable diseases and eighth among most burdensome diseases (Murray and Lopez, 2013). The one-year prevalence of migraine registered in the United States and Western Europe is 11% overall: 6% among men and 15–18% among women (Rasmussen and Olesen, 1992; Stewart et al., 1992). Chronic migraine, which is diagnosed when patients present 15 headache attacks or more per month over at least 3 consecutive months, affects about 1–2% of the general population (Lipton, 2011). After migraine has been diagnosed, its pharmacological treatment can either be abortive or prophylactic. Poor understanding of the mechanisms underlying migraine contributes to the current unsatisfactory prophylactic pharmacological treatment of migraine and particularly of chronic migraine. There are other rarer forms of primary headaches, primarily represented by cluster headache, which shows some similarities with migraine in terms of mechanisms and treatments.

Neurogenic inflammation

A subset of nociceptors is characterized by the ability of producing and releasing from their central and peripheral terminals the tachykinins, substance P (SP) and neurokinin A (NKA) and CGRP. Current neurochemical and functional identification of the subset of peptidergic somatosensory neurons has strengthened the seminal proposals of neurogenic vasodilatation by William Bayliss (1901) and of the ‘nocifensor system’ by Sir Thomas Lewis (Lewis, 1937). Indeed, some nerve endings of the C- and Aδ type of a subpopulation of nociceptors, when activated by noxious stimuli, orchestrate, via antidromic invasion of collateral fibres by propagated action potentials, an almost instantaneous defensive response, which encompasses a rapidly developing vasodilatation and plasma protein extravasation, all phenomena mediated by SP/NKA or CGRP liberated from their peripheral (perivascular) terminals. In experimental animals, and particularly in rodents, there is overwhelming evidence that plasma extravasation is mediated by SP or NKA which, by activating neurokinin-1 (NK1) receptors in endothelial cells of postcapillary venules, promotes the opening of gaps and the leakage of plasma proteins from the lumen to the interstitial space (Geppetti and Holzer, 1996; receptor nomenclature follows Alexander et al., 2013). In contrast, arteriolar vasodilatation, which is responsible for the neurogenic hyperaemic response, is mediated exclusively by the release of CGRP (Brain and Grant, 2004).

In the past, tachykinin release from perivascular nerve endings of trigeminal nociceptors and the ensuing meningeal plasma protein extravasation had been proposed as the underlying mechanism of migraine (Moskowitz et al., 1979). However, failure of NK1 receptor antagonists to ameliorate the headache and associated symptoms of migraine (Goldstein et al., 1997) definitively excluded any contribution of tachykinins, NK1 receptors and the plasma protein extravasation component of neurogenic inflammation to the disease. These negative clinical findings have underscored the difference between rodents and humans regarding the relative contribution of specific neurogenic inflammatory responses in the different species, and now the hypothesis that neurogenic inflammation contributes to migraine headaches has apparently been discarded.

CGRP and its distribution and function

However, basic and clinical investigation on CGRP has rejuvenated the seminal idea that neurogenic responses initiated by stimulation of trigeminal sensory nerve endings in cranial vessels are a major determinant of migraine pathogenesis (Moskowitz et al., 1979). CGRP, whose existence was predicted on the basis of the alternative splicing of the calcitonin gene (Amara et al., 1985), is, on a molar basis, one of the most powerful vasodilators known (Brain and Grant, 2004). Of the two forms of the peptide identified in humans, α-CGRP is mainly expressed in primary sensory neurons, while β-CGRP is primarily found in intrinsic enteric neurons (Brain and Grant, 2004). The α-CGRP isoform is a 37-amino acid peptide markedly expressed in sensory neurons of the dorsal root ganglia (DRG), trigeminal ganglia (TG) and vagal ganglia (VG; Amara et al., 1985; Brain and Grant, 2004). The mature peptide, CGRP, is then transported to the very terminal region of central and peripheral nerve endings, where it is stored in dense core vesicles to be secreted either in the dorsal spinal cord or in a variety of peripheral tissues, particularly around blood vessels (Geppetti and Holzer, 1996). Approximately 40–50% of TG neurons are CGRP-positive (Tajti et al., 1999b; Eftekhari et al., 2010) and a dense network of CGRP-positive nociceptors is present in rodent and human meningeal vessels (Tsai et al., 1988; Edvinsson et al., 1998b). CGRP is also expressed in specific areas of the CNS, including the hypothalamus, thalamus, periaqueductal gray, superior and inferior colliculi, amygdala, trigeminocervical complex (TCC), and the cerebellum (Hokfelt et al., 1992; van Rossum et al., 1997). Some of these brain areas may be relevant in migraine pathophysiology, as indicated by the ability of CGRP to alter synaptic and neuronal activity at the TCC and transmission of nociceptive signals to the thalamus and cortical regions (Storer et al., 2004; Goadsby, 2007). More recent findings have shown that CGRP receptor antagonists inhibit cortical spreading depression in a rat model (Tozzi et al., 2012). However, the precise role of CGRP in these central structures and the contribution of such effects to the migraine mechanism remain uncertain.

CGRP receptors

The functional CGRP receptor consists of a triad of proteins, comprising a classical GPCR, the calcitonin receptor-like receptor (CLR; Aiyar et al., 1996), a single transmembrane spanning protein called receptor activity-modifying protein 1 (RAMP1; McLatchie et al., 1998), required for the binding of CGRP to CLR and the receptor component protein (Ma et al., 2003) that characterizes the G-protein associated with the receptor. Components of the CGRP receptor complex are expressed in peripheral and central structures (Eftekhari and Edvinsson, 2010), such as cell bodies in TG, periaqueductal grey, and in the trigeminal nucleus caudalis (Oliver et al., 2002; Lennerz et al., 2008). However, it has not been convincingly established whether all these components assemble in a fully functional receptor at these anatomical sites. However, there is no doubt that vascular smooth muscle cells in arteries and arterioles, including those of the cranial circulation, express the entire and functional CGRP-receptor complex, as indicated by the robust vasodilatatory effect of CGRP(mediated by activation of adenylyl cyclase) in these vessels either in vitro and in vivo (Brain and Grant, 2004).

CGRP and migraine

While the anatomical site from which migraine attack originates and the initiating mechanism is still a mystery, the key role of CGRP in the migraine pathway is supported by a series of robust findings. After the in vivo preclinical observation that TG activation results in CGRP and SP release (Goadsby et al., 1988), the first indirect evidence suggesting a role of CGRP in migraine was obtained about 20 years ago, when it was shown that, during spontaneous migraine attacks, CGRP levels were elevated in samples of cranial venous blood (Goadsby et al., 1990). Increased CGRP levels have also been found in saliva during an acute migraine attack (Cady et al., 2009) or in cranial blood during nitroglycerine-evoked cluster headache attacks (Fanciullacci et al., 1995). Interestingly, successful treatment with antimigraine compounds, such as the triptans, is accompanied by a decrease in CGRP levels during migraine or cluster headache attacks (Fanciullacci et al., 1995; Goadsby and Hargreaves, 2000). It should be noted that others have failed to detect increased CGRP levels in cranial blood of migraineurs (Tvedskov et al., 2005). Another key observation, derived from a series of highly informative provocation experiments, showed that intravenous injection of CGRP induces migraine-like attacks in migraineurs (Lassen et al., 2002).

Conclusive evidence that CGRP plays a major role in migraine originated, however, from clinical trials that used various and chemically unrelated CGRP receptor antagonists, namely olcegepant (BIBN4096BS), telcagepant (MK-0974), MK-3207 and BI 44370 TA. Intravenous administration of 2.5 mg of olcegepant produced a response 2 h after treatment in 66% of treated patients, compared with a response rate of 27% observed in the placebo group (Olesen et al., 2004), a clinical response similar to that observed for triptans (Ferrari et al., 2002). This study made two important points; firstly it showed that blockade of CGRP receptors could be a valuable alternative mechanism for migraine treatment and secondly it provided the first clinical evidence of the involvement of CGRP in the mechanism(s) for migraine. The absence of direct vasoconstrictor activity and cardiovascular effects opened a novel scenario for and gave real hope of, CGRP receptor antagonists as an effective and well-tolerated migraine treatment (Petersen et al., 2005). Telcagepant was the first orally bioavailable CGRP receptor antagonist and, up to date, the most investigated compound of this new class of drugs. It showed efficacy in pain relief of migraine headache in phase II and III clinical trials (Ho et al., 2008a,b; Connor et al., 2009). Unfortunately, telcagepant increased plasma levels of liver transaminases in a few patients of a cohort of patients treated with the drug twice a day for 3 months (Han et al., 2010; Connor et al., 2011), and in patients affected by menstrual migraine enrolled in a short-term clinical study (Bigal et al., 2013). More recently, MK-3207, a new orally active CGRP receptor antagonist, demonstrated its superiority in comparison with placebo for migraine attack treatment (Hewitt et al., 2011). However, as for MK0974 (Merck & Co., 21 April 2009), MK3207 development was stopped due to the asymptomatic liver toxicity detected in some of the enrolled patients (Edvinsson and Linde, 2010). The development of both the oral CGRP antagonist, BI 44370 TA, which has been found superior to placebo end equieffective to eletriptan (Diener et al., 2011), and the orally bioavailable antagonist, BMS-927711 (Luo et al., 2012) has been stopped (Dolgin, 2013).

Studies with promising monoclonal antibodies (mAbs) against CGRP are ongoing (Bigal et al., 2013). At the time of this review, 4 mAbs are in clinical development for the treatment of migraine. Two of these mAbs, namely LY2951742 from Eli Lilly–Arteaus Therapeutics (Lilly Corporate Center, Indianapolis, IN, USA; Arteaus Therapeutics, Cambridge, MA, USA) and ALD403 from Alder Therapeutics (Alder Biopharmaceuticals, Inc., North Creek Parkway South Bothell, WA, USA), directly bind and neutralize CGRP. A third mAb, AMG 334 from Amgen Inc. (One Amgen Center Drive, Thousand Oaks, CA, USA), targets the CGRP receptor, while LBR-101 from Labrys Biologics–Pfizer (Labrys Biologics, Inc., San Mateo, CA, USA; Pfizer, New York, NY, USA) prevents the binding of CGRP to its receptor. Results of the early-phase clinical trials are not publicly available for any of these mAbs yet. According to data collected in preclinical studies, CGRP receptors are expressed on may different levels in the trigeminal vascular system of the cynomolgus monkey, such as the meningeal vasculature innervated by CGRP-positive nerve fibres, neurons and satellite cells in the trigeminal ganglion, and in the spinal trigeminal nucleus (Liu et al., 2011). Hence, in principle, mAbs targeting the CGRP system could exert their action at both the vascular and neuronal levels. Also, in the presence of a normal, un-compromised blood–brain barrier (BBB), about 0.1% of circulating IgG can enter the CNS, presumably through the circumventricular organs (Gu and Sigurdsson, 2011). Thus, even though mAbs could reach their target and prevent the interaction with the peptide and its receptors at central neuronal sites, it should be emphasized that the proportion of CGRP and /or CGRP receptors targeted by the minimal amount of mAbs that may cross the BBB is most likely negligible. Thus, blockade of the CGRP action at the perivascular level in intra- and extra-cranial vessels should currently remain the simplest and first explanation for any possible beneficial action of mAbs in migraine.

Despite the adverse reactions in the liver, which have not yet been conclusively ascribed to a class effect, the structural complexity of the CGRP receptor, which has represented the key hurdle for the development of small molecule antagonists (Moore and Salvatore, 2012), and the possible interaction of CGRP with different receptors (Walker and Hay, 2013), clinical data unequivocally demonstrate that, whatever the initiating mechanism(s) of the attacks, CGRP release and CGRP-receptor activation are major contributing mechanisms in migraine (Olesen et al., 2004; Ho et al., 2008a; Diener et al., 2011). Thus, identification of exogenous and endogenous stimuli that result in CGRP release from trigeminal neurons may be of paramount importance in decoding the molecular pathways that eventually cause or combine to worsen the headache and associated symptoms of migraine.

In the second part of this review, we first briefly present the transient receptor potential (TRP) channels and, in particular, those expressed by peptidergic primary sensory neurons, whose activation eventually results in neuropeptide release. We then summarize current knowledge regarding the role of one of these channels, the TRP ankyrin 1 (TRPA1), as a sensor of oxidative, nitrative and electrophilic stress and as a major player in various pain conditions, including migraine (see Table 1).

Table 1. Agents which trigger or inhibit migraine and cluster headache and act on TRP channels
TargetAgent (environmental, herbal, food, drug)Active compoundEffect on primary headacheAction on TRP channel
TRPA1Cigarette smoke

Crotonaldehyde acrolein formaldehyde nicotine


Migraine and cluster headache trigger (Rozen, 2010; Lima et al., 2011)Agonist (Bang et al., 2007a; McNamara et al., 2007; Andre et al., 2008; Talavera et al., 2009)
Tear gasO-chlorobenzylidene malononitrileHeadache trigger (Anderson et al., 1996)Agonist (Brone et al., 2008)
FormalinFormalinMigraine trigger (Wantke et al., 2000)Agonist (McNamara et al., 2007)
Umbellularia californicaUmbelluloneMigraine trigger (Immel, 2006)Agonist (Nassini et al., 2012a)

Tanacetum parthenium


ParthenolideMigraine preemptive Migraine abortive (Diener et al., 2005; Cady et al., 2011)Desensitizing agonist (Materazzi et al., 2013)
Angelica sinensisLigustilideMigraine preemptiveDesensitizing agonist (Zhong et al., 2011)
ParacetamolNAPQIMigraine abortiveDesensitizing agonist (Andersson et al., 2011)
Glyceryl trinitrateNOMigraine trigger (Iversen, 1995)Agonist (Miyamoto et al., 2009)
Ammonium chlorideAmmonium chlorideCluster headache trigger (Irlbacher and Meyer, 2002)Agonist (Fujita et al., 2008)
TRPV1Alcoholic beveragesEthanolMigraine trigger (Kelman, 2007)Agonist (Trevizani et al., 2002; Nicoletti et al., 2008)
CapsicumCapsaicinMigraine and cluster headache preemptive (Sicuteri et al., 1989; Fusco et al., 1994; 2003)Desensitizing agonist (Szallasi and Blumberg, 1999; O'Neill et al., 2012)
Multiple TRPTiger balmCamphorTension type headache abortive (Schattner and Randerson, 1996)

TRPV3 agonist (Moqrich et al., 2005)

TRPA1 antagonist (Sawada et al., 2008)

TRPV1 desensitizing agonist (Xu et al., 2005)

TRP channels

The TRP channel superfamily

From the original finding that vision in Drosophila melanogaster is produced by a mechanism that consists in an initial activation of a transient inward current associated with receptor stimulation (Minke, 1977; Montell, 1999), the identification of the larger family of the TRP channels has proceeded with an unprecedented pace. Currently, more than 50 members of the TRP family have been characterized (Nilius et al., 2007). Despite the wide heterogeneity of this family of ion channels, TRPs share a general role serving sensory transduction, because they contribute to vision, taste, olfaction, hearing, touch, and thermo- and osmo-sensation, making cells able to sense and respond to environmental changes.

TRP channels consist of six transmembrane domains (S1–S6) with both the amino (NH2) and carboxylic acid (COOH) termini localized to the cytosol. The COOH region is highly conserved among TRPs, whereas the NH2 region usually contains different ankyrin repeats, which consist of 33-residue motifs with a conserved backbone and variable residues that mediate specific protein–protein interactions (Sedgwick and Smerdon, 1999). The S1–S6 domains assemble as homo- or hetero-tetramers, with the pore domain formed by loops between S5 and S6, which permit a non-selective influx of cations. Although TRPs are described as non-selective Ca2+-permeable cation channels, their Ca2+/Na+ permeability ratio may vary markedly between different members of the superfamily and also among members of each subfamily (Nilius et al., 2007).

TRP channel gating may depend on direct activation of the channels by a plethora of physicochemical stimuli, including compounds of exogenous origin or endogenous signalling molecules (Nilius et al., 2007). TRP gating may also result from changes in the intracellular machinery as in the case of stimulation of the different isoforms of phospholipase C (PLC; Hardie and Minke, 1992; Niemeyer et al., 1996), following activation of GPCRs or tyrosine kinase receptors (Spehr et al., 2011). For instance, it has been hypothesized that PLC can modulate TRP channel activity through the hydrolysis of phosphatidylinositol (4,5) bisphosphate (PIP2), which leads to Ca2+ liberation from intracellular stores (Ramsey et al., 2006). Peculiar features of activation have been reported for certain TRP channels. These features may be exploited for a better understanding of how cells sense their surrounding environment, and may also represent the way to identify novel therapeutic targets.

In mammals, the TRP family consists of 28 proteins grouped into six subfamilies according to sequence identity and designated as TRP canonical (TRPC), TRP vanilloid (TRPV), TRP melastatin (TRPM), TRP polycystin (TRPP), TRP mucolipin (TRPML) and TRPA1 channels (Montell et al., 2002; Clapham, 2003). The mammalian TRPC subfamily comprises seven members (TRPC1–7), whose activation depends on the stimulation of GPCR and receptor tyrosine kinases (Montell, 1999), although TRPC1 channels seem to be directly activated by membrane stretch (Maroto et al., 2005). The TRPM comprises eight different members (TRPM1–8), which differently from TRPC and TRPV, do not contain ankyrin repeats within their NH2-terminal domain. Menthol and moderately low temperatures (<25°C) activate the TRPM8 channel. Both TRPP and TRPML families have been less extensively characterized. The TRPML family consists of three mammalian members (TRPML1–3). TRPML1 is widely expressed and appears to reside in late endosomes/lysosomes where it seems to act as a H+-sensitive channel to prevent overacidification (Soyombo et al., 2006). The heterogeneous TRPP family, consisting of three members according to structure, can be divided into PKD1-like (TRPP1-like) and PKD2-like (TRPP2-like) proteins (Hanaoka et al., 2000).

ThermoTRP channels in sensory neurons

Some TRP channels are abundantly expressed by the subsets of primary sensory neurons, which express neuropeptides. These include four of the six members of the TRPV subfamily, TRPV1, TRPV2, TRPV3 and TRPV4, characterized by their ability to sense warm–hot temperatures. TRPV channels also function as chemosensors for a number of naturally occurring (both exogenous and endogenous) and synthetic ligands. TRPV1, first identified as the receptor of the vanilloid compound, capsaicin, is responsive to high proton concentration (pH 6–5; Bevan and Geppetti, 1994; Tominaga et al., 1998), anandamide (Zygmunt et al., 1999) and a series of lipid derivatives (Ho et al., 2008c). Non-selective activators of TRPV3 and TRPV4 channels are camphor and hypotonic solutions respectively (Nilius et al., 2013). Synthetic agonists, such as the phorbol ester derivative 4α-phorbol 12,13-didecanoate (4α-PDD), low pH, citrate, endocannabinoids, arachidonic acid metabolites and NO may also activate TRPV3 channels (Watanabe et al., 2003; Vriens et al., 2005). Although the uricosuric agent probenecid has been identified as a selective stimulant of TRPV2 (Bang et al., 2007b), less information is available regarding the activators of TRPV2 channels. The TRPM8 channel is expressed in somatosensory neurons, but apparently not in those, which release tachykinins/CGRP (Bhattacharya et al., 2008). Finally, there is also evidence that TRPM3 channels, rather uniquely responsive to pregnenolone sulfate, seem to be localized to a subpopulation of primary sensory neurons (Wagner et al., 2008).

A large NH2-terminal with 17 predicted ankyrin repeat domains is the typical feature of the TRPA1 channel, the sole member of the TRPA subfamily. TRPA1 channels, first cloned from human foetal lung fibroblasts, are abundantly expressed in peptidergic nociceptors (Story et al., 2003; Bhattacharya et al., 2008), but these channels are also found in many non-neural cell types and tissues, including hair cells, pancreas, heart, brain, keratinocytes (Atoyan et al., 2009), urinary bladder (Streng et al., 2008), prostate gland (Gratzke et al., 2010), endothelium (Earley et al., 2009). and other vascular and perivascular cells (Earley, 2012), enterochromaffin cells (Nozawa et al., 2009), gastrointestinal tract (Izzo et al., 2012), odontoblasts (El Karim et al., 2010), dental pulp (El Karim et al., 2011), synovial fibroblasts (Kochukov et al., 2006), and epithelial and smooth muscle cells of the airways and lung (Nassini et al., 2012b). It has been extensively demonstrated that TRPA1 channels play a key role in the detection of pungent or irritant principles, including compounds contained in various spicy foods, such as allyl isothiocyanate (mustard oil) contained in horseradish (Jordt et al., 2004), allicin and diallyldisulfide contained in garlic (Bautista et al., 2005), cinnamaldehyde contained in cinnamon (Bandell et al., 2004), and capsiate (Shintaku et al., 2012). Additional spices or food ingredients that may activate TRPA1 channels are gingerol (in ginger), eugenol (in cloves), methyl salicylate (in wintergreen), carvacrol (in oregano), and thymol (in thyme and oregano; Bandell et al., 2004; Xu et al., 2006; Lee et al., 2008). Environmental irritants and industry pollutants, such as acetaldehyde, formaldehyde, hydrogen peroxide, hypochlorite, isocyanates, ozone, carbon dioxide, ultraviolet light and acrolein, a highly reactive α,β-unsatured aldehyde present in tear gas, cigarette smoke, smoke from burning vegetation and vehicle exhaust, and hydrogen sulfide (H2S) also activate TRPA1 channels (Bautista et al., 2006; Bang et al., 2007a; McNamara et al., 2007; Andersson et al., 2008; Bessac et al., 2008; 2009; Sawada et al., 2008; Hill and Schaefer, 2009; Taylor-Clark and Undem, 2010; Wang et al., 2010; Miyamoto et al., 2011). Recently, it has been reported the ability of cannabichromene, a non-psychotropic Cannabis-derived cannabinoid with anti-inflammatory (Romano et al., 2013) and analgesic properties (Maione et al., 2011) to activate TRPA1 channels (De Petrocellis et al., 2011). It has also been proposed that these channels function as a detector of mechanical stimuli and noxious cold (<17°C; Story et al., 2003), although these proposals remain controversial (Jordt et al., 2004; Latorre, 2009). TRPV1, TRPV2, TRPV3, TRPV4, TRPA1 and TRPM8 have been collectively labelled as thermoTRP channels because they can be activated by a large range of temperatures from noxious cold to noxious heat (Vay et al., 2012).

Members of the TRP family expressed in sensory neurons are primarily involved in the detection of noxious physical (thermal and mechanical) and chemical stimuli. Among the 6–7 TRPs expressed by nociceptors, recent pathophysiological and pharmacological findings have pointed to TRPV1 and TRPA1 channels as main contributors in models of inflammatory and neuropathic pain (Fernandes et al., 2012). It should be underlined that TRPV1 and TRPA1 channels co-localize in a subpopulation of non-myelinated or thinly myelinated C- or Aδ-fibre neurons of the DRG, TG and VG. The population of TRPV1-positive neurons seems to be larger than the TRPA1-positive subpopulation (Story et al., 2003; Bhattacharya et al., 2008). Both TRPV1 and TRPA1 channels coexist within neuropeptides (SP, NKA and CGRP) in the same nociceptive neurons. While TRPA1 expression seems to be confined to the peptidergic neuronal subpopulation, TRPV1-positive neurons appear to be also non-peptidergic (Story et al., 2003; Bhattacharya et al., 2008). However, more recent studies have identified subpopulations of DRG neurons that co-stain for IB4 (isolectin B4, a marker of non-peptidergic neurons) or the purinergic P2X3 receptor (also expressed by non-peptidergic neurons) and TRPA1 channels (Kim et al., 2010; Barabas et al., 2012).

Mutation of TRP channels has been linked to various diseases affecting different organs or systems. However, only a small number of TRP channelopathies, also known as ‘TRPpathies’, has been definitively identified so far. TRPpathies encompass neurological disorders, including scapuloperoneal hereditary motor neuropathy and Charcot–Marie–Tooth disease type-2C, due to TRPV4 mutation or Guamanian amyotrophic lateral sclerosis/parkinsonism dementia complex, due to TRPM2 or TRPM7 mutation, renal diseases, including focal segmental glomerulosclerosis, due to TRPC6 mutation or autosomal dominant polycystic kidney disease, due to TRPP2 mutation, and complex skeletal dysplasias, including brachyolmia type 3, spondylometaphyseal dysplasia, Kozlowski type and autosomal dominant metatropic dysplasia due to TRPV4 mutation, among others (Nilius and Owsianik, 2010). Of interest for the present discussion, recently, a familial episodic pain syndrome has been attributed to a gain of function mutation of TRPA1 channels (Kremeyer et al., 2010).

TRPA1 channels, oxidative stress and nitrative stress in pain and inflammation

In aerobic organisms, the balance between oxidation and reduction (redox state) is of paramount importance for physiological homeostasis. The conservation of an optimal redox state is pursued through diverse mechanisms also involving the transformation in reactive oxygen species (ROS), including O2–, OH, H2O2 and O3, which affect cellular functions deeply, but can also lead to irreversible damage, up to cell death. Oxidative stress has been claimed as a mechanism for a number of diseases, including inflammatory pain, neuropathic pain and migraine. An unprecedented series of findings has shown that TRPA1 channels are activated by ROS, reactive nitrogen species (RNS) and other electrophiles, thus identifying the channel as a sensor of oxidative and nitrative stress generated at sites of inflammation or tissue injury. Indeed, Ca2+ influx and activation of membrane currents in sensory neurons induced by ROS, RNS or RCS (reactive carbonyl species) are absent in TRPA1−/− mice and are blocked by TRPA1 channel antagonists (Bautista et al., 2006; Trevisani et al., 2007; Bessac et al., 2008; Taylor-Clark and Undem, 2010). Activation by hyperoxia (86% O2) suggests that TRPA1 channels function as sensors, exclusively for abnormal redox states (Takahashi et al., 2011). α,β-Unsaturated aldehydes produced by membrane phospholipid peroxidation by ROS, such as acrolein (Bautista et al., 2006), 4-hydroxynonenal (Trevisani et al., 2007) or oxononenal (Taylor-Clark et al., 2008), or other by-products of oxidative stress, such as hydrogen peroxide (Bessac et al., 2008), hypochlorite (Bessac et al., 2008), or nitrative stress by-products, such as nitroleic acid or NO (Taylor-Clark et al., 2009) and other reactive molecules, all share the ability to activate TRPA1 because of their reactive properties, which result in the carbonylation, nitrosilation or oxidation of specific cysteine (C619, C639, C663, C415, C422 and C622) or lysine (K708) residues (Hinman et al., 2006; Macpherson et al., 2007).

Oxidative/nitrative stress and the ensuing TRPA1 activation result from endogenous processes driven by inflammatory or degenerative conditions, but may also derive from exogenous stimuli, which per se are already suited for TRPA1 channel stimulation. Thus, crotonaldehyde, acrolein (Andre et al., 2008), acetaldehyde (Bang et al., 2007a) and possibly other reactive molecules and nicotine (Talavera et al., 2009), all contained in cigarette smoke, gate the TRPA1 channels on airway sensory nerve terminals to release tachykinins and CGRP, which in turn mediate the early inflammatory response that follows acute exposure to cigarette smoke (Andre et al., 2008). Certain volatile anaesthetics, including the irritants isoflurane and desflurane, during induction and emergence from anaesthesia cause a strong cough reflex that can precipitate laryngospasm, a potentially life-threatening complication. Their ability to gate TRPA1 channels, thus producing sensory nerve activation and neurogenic inflammation responses, may be the underlying mechanism for such an adverse reaction (Matta et al., 2008; Eilers et al., 2010). Chemotherapeutic agents are known to produce remarkable oxidative stress, which is likely to contribute to their anticancer activity, even if this issue remains highly debated (Saeidnia and Abdollahi, 2013).

However, oxidative stress seems to be responsible for serious adverse events, including chemotherapeutic-induced peripheral neuropathy (CIPN), which characterized by paraesthesias, spontaneous pain, and typically by prolonged mechanical and cold hypersensitivity, reduces quality of life, and often causes hospitalization and therapy discontinuation (Cavaletti and Marmiroli, 2010). We recently found in a mouse model of CIPN that oxaliplatin, paclitaxel or bortezomib (Nassini et al., 2011; Materazzi et al., 2012; Trevisan et al., 2013) produce a prolonged condition of mechanical and cold hypersensitivity that lasts for 11–15 days. When established, the hypersensitivity is completely, although transiently, reverted by a ROS scavenger (α-lipoic acid) or a TRPA1 channel antagonist (HC-030031). However, in TRPA1-deleted mice, or if the ROS scavenger or the TRPA1 antagonist was given just before and for 6 h after bortezomib or oxaliplatin administration, the mice were completely and permanently protected from the development of the hypersensitivity (Trevisan et al., 2013). Because a marker of oxidative stress was transiently increased in mouse plasma within the first few hours after anticancer drug administration, the hypothesis was advanced that sensitization of TRPA1 channels by oxidative stress is required to establish and maintain the prolonged hypersensitivity by chemotherapeutic agents (Trevisan et al., 2013). These channels also contribute to mechanical hyperalgesia in models of inflammatory pain, such as those induced by carrageenan (Moilanen et al., 2012), complete Freund's adjuvant (da Costa et al., 2010), and in a caerulein-induced model of pancreatitis (Schwartz et al., 2013). However, in these cases, the association of TRPA1 channel activation/sensitization and oxidative stress has not been investigated.

TRPA1 channels in pain and migraine

Anecdotal reports and epidemiological findings indicate that a large series of endogenous or exogenous agents trigger headache in migraineurs (Courteau et al., 1994; Kelman, 2007; Lima et al., 2011). In the context of this review, it is important to emphasize that, among the highly heterogeneous series of triggering factors, certain foods or exposure to environmental agents provoke migraine headaches (Courteau et al., 1994; Kelman, 2007; Lima et al., 2011). In terms of foods, 40–50% of migraineurs are sensitive to alcohol or chocolate (Kelman, 2007). Regarding environmental agents, there is evidence suggesting that migraine and cluster headache are favoured in susceptible individuals by increased concentration of air pollutants, as indicated by a greater number of emergency room visits for headache under such circumstances (Szyszkowicz, 2008). Furthermore, exposure to tear gas induces a series of toxic effects such as cough, chest pain, dyspnoea and headache (Anderson et al., 1996). Cigarette smoke inhalation affects headache occurrence in migraineurs (Lima et al., 2011) and prolonged or repeated exposure to cigarette smoke may increase cluster headache frequency (Rozen, 2010). However, a mechanistic explanation for the ability of this apparently unrelated series of substances to trigger migraine headaches is still missing.

The α,β-unsaturated aldehyde, acrolein, contained in cigarette smoke (Bautista et al., 2006), is produced endogenously by plasma membrane peroxidation by oxidative stress. Acrolein, when acting on sensory nerve endings, causes neurogenic inflammation, by a mechanism dependent on capsaicin-sensitive peptidergic primary afferent fibres, via activation of TRPA1 channels (Bautista et al., 2006; Geppetti et al., 2008). Accordingly, compounds contained in cigarette smoke, including acrolein and crotonaldehyde, produce airway inflammatory responses through stimulation of TRPA1 channels expressed by vagal sensory nerve endings (Andre et al., 2008). Interestingly, acrolein, when applied to the nasal mucosa of rats, enhances the meningeal blood flow by a mechanism dependent on TRPA1 channel activation and the subsequent release of CGRP (Kunkler et al., 2011). Thus, it could be hypothesized that acrolein, because of its ability to activate these channels and thereby produce CGRP release, mediates neurogenic inflammation and headache induced by toxic environmental irritants, including cigarette smoke inhalation (Geppetti et al., 2008; Kunkler et al., 2011). The same mechanism could also be hypothesized for tear gases, in particular for 2-chlorobenzalmalononitrile (CS tear gas), which has shown to be able to induce headache (Anderson et al., 1996), and with similar constituents of tear gases, is one of the most potent and specific TRPA1 channel agonists characterized so far (Brone et al., 2008). Although the effect of H2S has not often been reported, intoxication with H2S has been described to trigger headache attacks (Hirsch and Zavala, 1999). Underestimation of the phenomenon may be due to the fact that the intoxication is frequently associated with loss of consciousness of the intoxicated subjects. The discovery that H2S stimulates TRPA1 channels (Miyamoto et al., 2011; Okubo et al., 2012) and that this activation results in CGRP release (Pozsgai et al., 2012; White et al., 2013) suggests a mechanistic pathway for H2S-induced headache.

Knowledge and use of herbal medicines, as often happens in biomedical investigation, may help to elucidate mechanisms of disease, to validate therapeutic targets, and eventually to direct pharmaceutical development. The California bay laurel, Umbellularia californica, is also known as the ‘headache tree’ because of the ability of its scent to cause headache attacks (Immel, 2006). Recent evidence showed that exposure to the scent of Umbellularia californica triggers cluster headache attacks (Benemei et al., 2010). Umbellulone, is a major volatile component of Umbellularia californica, with a specific chemical feature, resembling those necessary for TRPA1 agonism. Although umbellulone possesses a β,β-dialkyl substitution, which should inhibit reaction with thiol groups (LoPachin et al., 2008), it rather surprisingly reacts in a ‘click-fashion’ with the biogenic thiol cysteamine, producing a Michael adduct (Nassini et al., 2012a). As the closely related, but not irritant, terpenoid enones, (+)-verbenone and (+)-piperitone, do not produce such a chemical reaction, Michael acceptor behaviour seems to be required for the sensory noxious activity of umbellulone. Umbellulone, most likely through this chemical property, caused TRPA1-dependent CGRP release in vitro and nociceptive behaviour in vivo in rats and mice. Finally, and more importantly, when applied intranasally (likewise in the acrolein experiments), umbellulone evoked TRPA1-mediated and CGRP-dependent neurogenic meningeal vasodilation (Nassini et al., 2012a). Although a reflex pathway, originating in the nasal cavity, and resulting in CGRP-dependent meningeal vasodilatation, has been proposed to explain the vascular action of acrolein and umbellulone (Kunkler et al., 2011; Nassini et al., 2012a), the precise mechanism involved in such a response remains to be elucidated. There is evidence that TRPA1 channels are expressed by endothelial cells of rat cerebral and cerebellar pial arteries (but not in endothelium of other vascular beds), where they mediate endothelium-dependent vasodilation (Earley et al., 2009). However, a contribution of such CGRP-independent mechanism for umbellulone- and acrolein-evoked vasodilation is unlikely, as the response in meningeal arteries by the two TRPA1 channel agonists was completely abolished by CGRP receptor antagonism (Kunkler et al., 2011; Nassini et al., 2012a).

A series of substances, now identified as TRPA1 channel agonists, has been reported in the past to cause migraine or non-migraine headaches after inhalation by migraine patients or non-migraine individuals respectively (Courteau et al., 1994; Kelman, 2007; Lima et al., 2011). Ammonium chloride is an agonist for these channels (Bessac and Jordt, 2010) and its inhalation has been reported to trigger cluster-like headache attacks (Irlbacher and Meyer, 2002). Similarly, formalin has long been known as a headache-producing agent (Wantke et al., 2000), and it has recently been recognized as a TRPA1 agonist (McNamara et al., 2007).

The contribution of herbal medicines to the understanding of the role of TRPA1 channels in headache mechanism is not limited to the irritant substance, umbellulone. Very recently, it has been demonstrated that parthenolide, a bioactive compound contained in the antimigraine preparations from Tanacetum parthenium (also known as feverfew), acts as a partial agonist at TRPA1 channels (Materazzi et al., 2013). In addition, parthenolide, after an initial and moderate activation, produces a profound and persistent desensitization of these channels and a complete defunctionalization of sensory neurons (including meningeal trigeminal nerve terminals), which are rendered unable to release CGRP upon exposure to any stimulus (Materazzi et al., 2013). These findings suggest that the antimigraine action of preparations of feverfew, either used as a pre-emptive treatment (Diener et al., 2005) or as an acute medication (Cady et al., 2011), may be due to the ability of parthenolide to deactivate the trigeminovascular-CGRP system. The desensitizing action of parthenolide does not seem unique to this molecule, as ligustilide (Zhong et al., 2011), has recently been identified as a TRPA1 channel agonist, which may induce TRPA1 and sensory neuron desensitization. Ligustilide, contained in elevated concentrations in herbal remedies, is used in Chinese and North American traditional medicine to treat pain and headaches (Li et al., 2011).

Some medicines may also provoke migraine. Nitroglycerine, a NO donor with a direct vasodilator action on vascular smooth muscle, has been known for a long time as a prototypical headache-causing agent (Thomsen and Olesen, 2001) and as a valuable experimental tool to provoke migraine-like attacks (Iversen, 1995). One possible explanation for the pro-migraine action of nitroglycerine is based on the ability of NO to produce vasodilatation of cranial arteries (Shevel, 2011). This view is supported by pharmacological evidence that nitroglycerine-induced migraine-like headaches are reversed by sumatriptan (Iversen and Olesen, 1993; 1996), presumably by its direct vasoconstrictor action (Asghar et al., 2011; Amin et al., 2013), but not by a CGRP receptor antagonist (Tvedskov et al., 2010). NO may also release CGRP in vitro (Wei et al., 1992) and in vivo (Fanciullacci et al., 1995), even if this effect of NO has not always been confirmed (Eltorp et al., 2000). Although NO stimulates sensory neurons by activating TRPA1 channels (Miyamoto et al., 2009), it is not known whether NO releases CGRP from trigeminal neurons via TRPA1 stimulation. It should be, however, underlined that vasodilation in vivo (Iversen, 1995) and CGRP release in vitro (Wei et al., 1992) are very early phenomena, which occur within a few minutes after drug exposure. In contrast, clinical investigation with nitroglycerine has revealed a stereotyped and delayed time-course of the migraine-like pain that develop only 4–5 h after (Iversen, 1995) drug administration. Thus, a temporal mismatch exists between vasodilation/acute CGRP release and the occurrence of the migraine headache evoked by NO. This represents a major reason for criticism of the hypothesis that these early neurovascular actions are responsible for the pro-migraine action of the drug. Changes in neuronal sensitivity, driven by the exposure to nitroglycerine/NO, which requires 4–5 h to fully exhibit their pro-migraine potential, are more likely to be responsible for the phenomenon. However, the possible role of TRPA1 in this key process has not yet been explored.

Finally, TRPA1 channel research has provided new insights in the analgesic activity of an old medicine. The paracetamol (acetaminophen) metabolite, N-acetyl-p-benzoquinone imine (NAPQI), activated the TRPA1 channel and thereby evoked a moderate and reversible neurogenic inflammatory response, which may have contributed to airway inflammation (Nassini et al., 2010) and thus could favour asthma in susceptible individuals (Beasley et al., 2008). These initial results were followed by further data, relevant to understanding the hitherto unexplained analgesic action of paracetamol. NAPQI potentially generated by autochthone cytochrome activity within the spinal cord desensitized TRPA1 channels, which induced channel-dependent, anti-nociceptive actions (Andersson et al., 2011). However, it is not known whether this novel spinal mechanism mediated by inhibition of TRPA1 channels, possibly relevant for the general analgesic actions of paracetamol, could be of interest for the antimigraine activity of the drug.

Other TRPs in migraine and sensitization processes

TRPV1, probably because of the long-standing use of capsaicin as a pain remedy, was the first TRP channel studied in migraine. TRPV1 is a multifunctional channel involved in thermo- (heat) and chemo-sensation, functioning as a receptor for a number of seemingly unrelated noxious stimuli. The unique pharmacological property of capsaicin is based on an initial transient TRPV1 activation, associated in vivo to an irritating and painful sensation, which is quite rapidly followed by a durable refractory state in which nociceptors do not respond to subsequent challenges with capsaicin or any other irritant/painful stimulus. The marked desensitization produced by capsaicin accounts for the widespread use of capsaicin skin creams, ointments, patches and other preparations for the treatment of localized neuropathic pain (Backonja et al., 2008). A similar desensitizing property that follows site-specific injections of the capsaicin analogue and ultrapotent TRPV1 channel agonist, resiniferatoxin (Szallasi and Blumberg, 1993), has been successfully used in bladder disorders (Lazzeri et al., 2000; 2004; Peng and Kuo, 2007), and is being evaluated as a long-lasting analgesic treatment in cancer patients with refractory chronic pain (NCT00804154). Based on the assumption that this procedure, which targets the capsaicin ‘receptor’ could block meningeal afferents of the first branch of the trigeminal nerve, topical capsaicin application has been applied to the nasal mucosa to prevent cluster headache (Sicuteri et al., 1989) or migraine (Saper et al., 2002; Fusco et al., 2003) attacks. Despite the fact that results of clinical studies with intranasally delivered formulations of both capsaicin and capsaicin analogues (e.g. civamide) have often been compromised by methodological limits and small sample size (Saper et al., 2002; Fusco et al., 2003), a recent review on the treatment of cluster headache proposed a level B recommendation for intranasal civamide for prophylactic purposes (Francis et al., 2010).

TRPV1 channels under certain circumstances could also play a pathophysiological role in migraine. Alcoholic beverages are reported to induce migraine in about 40% of the patients (Kelman, 2007), but the reason for such susceptibility is unknown. Ethanol acts as a TRPV1 channel stimulant, probably because it reduces the threshold temperature for channel activation by 8°C, from 42/43 to 35°C, which is below the normal body temperature (Trevisani et al., 2002). In addition, and importantly for migraine pathophysiology, ethanol, by targeting TRPV1 channels, evokes CGRP release and the consequent vasodilation in meningeal vessels (Nicoletti et al., 2008). The ability of ethanol to sensitize TRPV1 channels is not limited to temperature, as TRPV1-mediated responses to anandamide and protons are also exaggerated by about 10- and 50-fold respectively (Trevisani et al., 2002). Thus, it may be hypothesized that the ability of alcoholic beverages to trigger migraine depends on individual susceptibility as well as the presence of co-factors, which potentiate ethanol or temperature-gating actions on TRPV1 channels.

The TRPV4 channel is activated by hypoosmolar solutions and membrane stretching and is expressed in peptidergic sensory neurons (Vergnolle et al., 2010). This feature could be of interest in migraine, the pain of which is often described as throbbing. It has also been reported that pregnenolone sulfate is a TRPM3 channel agonist in DRG neurons (Vriens et al., 2011). This finding seems of particular interest, as menstrual cycle, pregnancy and menopause are major determinants of the frequency and severity of migraine attacks. It also suggests a possible direct mechanism that may explain the strong association between female sex hormones and sensory neuron activation in migraine. However, no studies have yet addressed the role of either TRPV4 or TRPM3 channels in migraine or associated headaches.

Chronic migraine and medication overuse headache (MOH, chronic headache accompanied by overuse of symptomatic medicines) are conditions underlined by a progressively worsening hypersensitivity to a host of usually innocuous stimuli. Thus, allodynia represents a characteristic of chronic migraine and MOH (Schurks and Diener, 2008; De Felice et al., 2011). Hypersensitivity may also develop during each individual attack, where sensitization of trigeminal neurons is considered a first step, responsible for the perception of headache throbbing pain (Strassman and Levy, 2006), of a more complex process, which, involving second-order neurons, contributes to cephalic allodynia and muscle tenderness (Burstein et al., 2000; 2010). Hyperproduction of endogenous inflammatory mediators may take part in the initial phenomenon, whereas a central mechanism may activate and sensitize thalamic trigeminovascular neurons (Noseda and Burstein, 2013). A number of neurotransmitters and neural circuitries may contribute to aggravate sensitization in each individual attack, or promote the transition of migraine from an episodic to a chronic condition. There is abundant evidence that TRP channels are sensitized following exposure to proinflammatory mediators, and by intracellular pathways (Trevisani et al., 2002; Nilius et al., 2007; Nilius and Owsianik, 2010; Selescu et al., 2013; Trevisan et al., 2013). However, specific information on the role of TRP channels, and in particular TRPV1 and TRPA1, in the sensitization of trigeminal afferents and/or central nociceptive neurons, has not yet been addressed.


Genetic investigation has provided little help to identify the pathophysiological features that determine migraine susceptibility. Similarly, neuroimaging or biomolecular studies have not yet disclosed the underlying anatomical and neurochemical pathways that result in the migraine attack. In contrast, pharmacological studies and clinical trials have made fundamental contributions to establishing some key players in migraine headache. The ability of non-steroidal anti-inflammatory drugs (NSAIDs) to treat migraine indicates that prostanoids have a role in the disease. However, their action is far from specific, as NSAIDs are used in practically all types of pain. Triptans are of paramount importance in the acute relief of migraine attack. However, as they are receptor agonists (do not block any endogenous agent), and because of their multiple pharmacological actions (vasoconstrictors, inhibitors of CGRP release from perivascular sensory nerve terminals, inhibitors of neuronal transmission within central brain areas), they are not better suited to elucidate the underlying mechanisms of migraine. In contrast, the ability of CGRP receptor antagonists to abolish the pain and associated symptoms of migraine attack, coupled to the lack of evidence that CGRP plays any major role in other types of pain, strongly and specifically implicates CGRP in migraine. Thus, knowledge of the mechanisms that directly cause, or regulate, after a process of sensitization or desensitization, CGRP release from nerves is of paramount importance for our understanding of migraine pathophysiology. TRP channels expressed in peptidergic nociceptors are among the molecular entities that encompass all the features required for this type of investigation. TRPA1 channels, which are activated or sensitized by some known triggers of migraine, along with a wide variety of other stimuli, and are inhibited by analgesic and antimigraine medicines (see Figure 1), would represent a major target for novel drugs to treat migraine and the associated primary headaches.

Figure 1.

Schematic representation of the activity of several agents (drugs, herbal medicines, endogenous and exogenous compounds), which, by targeting the TRPA1 channel, may positively or negatively affect the migraine attack via the release of CGRP and SP from peripheral and central endings of trigeminal neurons. (A) Some agents may behave as partial agonists or after an initial activation may lead to a profound and enduring channel desensitization. Both mechanisms by inhibiting CGRP release may eventually ameliorate migraine and cluster headache attacks. (B) In contrast, agonists of the TRPA1, by channel stimulation and the ensuing release of neuropeptide, may trigger migraine and cluster headache attacks.

Conflict of interest

The authors state no conflict of interest.