Brain Research Reviews 48 (2005) 438 – 456
`
`Review
`
`www.elsevier.com/locate/brainresrev
`
`Neurobiology in primary headaches
`
`Lars Edvinssona,*, Rolf Uddmanb
`
`aDepartment of Internal Medicine, University Hospital, S-221 85 Lund, Sweden
`bDepartment of Otorhinolaryngology, Malmo¨ University Hospital, Malmo¨ , Sweden
`
`Accepted 8 September 2004
`Available online 18 November 2004
`
`Abstract
`
`Primary headaches such as migraine and cluster headache are neurovascular disorders. Migraine is a painful, incapacitating disease that
`affects a large portion of the adult population with a substantial economic burden on society. The disorder is characterised by recurrent
`unilateral headaches, usually accompanied by nausea, vomiting, photophobia and/or phonophobia. A number of hypothesis have emerged to
`explain the specific causes of migraine. Current theories suggest that the initiation of a migraine attack involves a primary central nervous
`system (CNS) event. It has been suggested that a mutation in a calcium gene channel renders the individual more sensitive to environmental
`factors, resulting in a wave of cortical spreading depression when the attack is initiated. Genetically, migraine is a complex familial disorder
`in which the severity and the susceptibility of individuals are most likely governed by several genes that vary between families. Genom wide
`scans have been performed in migraine with susceptibility regions on several chromosomes some are associated with altered calcium channel
`function. With positron emission tomography (PET), a migraine active region has been pointed out in the brainstem. In cluster headache, PET
`studies have implicated a specific active locus in the posterior hypothalamus. Both migraine and cluster headache involve activation of the
`trigeminovascular system. In support, there is a clear association between the head pain and the release of the neuropeptide calcitonin gene-
`related peptide (CGRP) from the trigeminovascular system. In cluster headache there is, in addition, release of the parasympathetic
`neuropeptide vasoactive intestinal peptide (VIP) that is coupled to facial vasomotor symptoms. Triptan administration, activating the 5-HT1B/
`1D receptors, causes the headache to subside and the levels of neuropeptides to normalise, in part through presynaptic inhibition of the cranial
`sensory nerves. These data suggest a central role for sensory and parasympathetic mechanisms in the pathophysiology of primary headaches.
`The positive clinical trial with a CGRP receptor antagonist offers a new promising way of treatment.
`D 2004 Elsevier B.V. All rights reserved.
`
`Theme: Sensory system
`Topic: Pain modulation: anatomy and physiology
`
`Keywords: Migraine; Cluster headache; CGRP; VIP; Trigeminovascular reflex; Autonomic nerves
`
`Contents
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`Introduction .
`1.
`.
`.
`.
`.
`.
`.
`.
`2. Where does the attack start? .
`.
`.
`.
`.
`.
`.
`.
`3.
`The ion channel connection .
`4. Nerves in the walls of intracranial vessels .
`4.1.
`Sympathetic nervous system .
`.
`.
`.
`4.2.
`Parasympathetic nervous system .
`.
`4.3.
`Sensory nervous system .
`.
`.
`.
`.
`.
`4.4.
`Intracerebral innervation .
`.
`.
`.
`.
`.
`5. Neurotransmitters in primary headaches .
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`439
`439
`441
`442
`442
`443
`444
`445
`446
`
`* Corresponding author. Tel.: +46 46 17 14 84; fax: +46 46 18 47 92.
`E-mail address: lars.edvinsson@med.lu.se (L. Edvinsson).
`
`0165-0173/$ - see front matter D 2004 Elsevier B.V. All rights reserved.
`doi:10.1016/j.brainresrev.2004.09.007
`
`1
`
`EX2155
`Eli Lilly & Co. v. Teva Pharms. Int'l GMBH
`IPR2018-01427
`
`

`

`L. Edvinsson, R. Uddman / Brain Research Reviews 48 (2005) 438–456
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`Trigeminal ganglion stimulation .
`5.1.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`5.2. Migraine attacks.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`5.3. Cluster headache .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`5.4.
`Trigeminal neuralgia .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`5.5. Chronic paroxysmal hemicrania (CPH) .
`.
`.
`.
`.
`.
`.
`5.6.
`Tension-type headache .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`6. Central mechanisms in headache .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`6.1. How is the trigeminovascular reflex initiated? .
`6.2. What is the role of the trigeminocervical complex?.
`7. Central sensitization .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`8.
`Peripheral sensitization .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`9.
`Summary .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`Acknowledgments .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`References .
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`
`439
`
`446
`446
`447
`447
`447
`448
`448
`448
`448
`449
`449
`450
`451
`451
`
`1. Introduction
`
`The primary headaches include migraine, tension-type
`headache (TTH), cluster headache, other trigeminal auto-
`nomic cephalalgias and other headaches [147]. Tension-type
`headache is the most common of these in the general
`population, however, since little data exist for a neuro-
`vascular component, we have only described those briefly
`below [5]. Migraine headaches are ascribed as neuro-
`vascular disorders which world-wide afflict up to 15–20%
`of the general population. The socio-economic implications
`are extensive with considerable impact on productivity and
`quality of life. In Europe alone, it is calculated that 600,000
`days of work are lost daily. Migraine, which is the most
`common type, is characterised by attacks of moderate to
`severe headache that
`last for 4–72 h, often unilateral,
`pulsating and associated with photophobia/phonophobia
`and/or nausea/vomiting [146]. In migraine with aura, the
`headache is preceded by transient
`focal neurological
`symptoms, most often contralaterally [84].
`Cluster headache is another of the primary headaches; it
`has a distinct clinic with devastating pain. Some of the
`features of cluster headache overlap with those of other
`primary vascular headaches. The pain usually occurs around
`the eye and is described as retro-orbital or temporal. This
`implies involvement of the ophthalmic (first) division of the
`trigeminal nerve. In addition to the pain, there are signs of
`parasympathetic overactivity, e.g., lacrimation, nasal con-
`gestion and injection of the eye. Short-lasting headaches
`associated with autonomic symptoms may sometimes be
`confused with cluster headache. Although the exact causes
`of the primary headaches remain unknown, some pieces of
`the pathophysiological puzzle are starting to fall into place,
`particularly after a series of elegant positron emission
`tomography (PET) studies [132–134]. During the last 20
`years, there has been a heated debate whether the primary
`headaches are neurogenic or vascular in origin. However,
`current molecular and functional studies suggest a way to
`incorporate the different aspects into an integrated hypoth-
`esis as neurovascular headaches [39,84,156].
`
`In susceptible individuals, changes in environmental or
`physiological states are known to trigger the migraine
`headache. Migraine susceptibility has been linked to
`mechanisms regulating central sensitization. The systems
`that govern neuronal excitability involve homeostatic
`mechanisms and intracellular signalling pathways. The
`demonstration that mutations in the calcium channel gene
`CACNA1A, in approximately 50% of families suffering
`from the rare and severe familial hemiplegic migraine
`(FHM), has offered some hope that there is a molecular
`genetic cause also of the more common types of migraine
`[150,188]. However, it is well recognised that the central
`nervous system (CNS) is devoid of sensory pain receptors,
`and intracranially, it is only blood vessels in the dura and the
`circle of Willis that are supplied with sensory nerves and
`receptors that can respond to thermal, mechanical or
`distensional stimuli [147,159].
`
`2. Where does the attack start?
`
`Some researchers have suggested that migraine is a
`disease comprised of two main subtypes, migraine with aura
`and migraine without aura. In the former,
`the aura is
`characterized most often by visual field disturbances, but
`sometimes also by additional somatosensory disturbances.
`In these patients, changes in cortical blood flow correlate
`with areas of hypoperfusion, but no subsequent spreading
`from the area of hypoperfusion can be demonstrated,
`possibly because these patients have been studied late
`during the attacks. Olesen et al. [144] were the first to
`observe in patients examined early at the onset of induced
`migraine attacks, a pattern of localized blood flow decrease
`that spread contiguously over the cerebral cortex. This
`pattern of bspreading oligemiaQ or bspreading hypo-
`perfusionQ was apparent only in patients who had migraine
`with aura. The hypoperfusion was ipsilateral to the headache
`and contralateral to the symptoms of the aura. In one subject
`who suffered a migraine attack during a series of cerebral
`blood flow measurements with PET [201], the headache
`
`2
`
`

`

`440
`
`L. Edvinsson, R. Uddman / Brain Research Reviews 48 (2005) 438–456
`
`was associated with bilateral hypoperfusion which started in
`the occipital lobes and spread anteriorly into the temporal
`and parietal lobes. This provided high-resolution evidence
`of the spreading nature of the hypoperfusion associated with
`a spontaneous migraine attack. This view was further
`supported by a study of blood oxygenation level dependent
`(BOLD) signal changes reflecting the balance between
`oxygen delivery and oxygen consumption. In one patient,
`two attacks of induced migraine aura showed an increase in
`the mean magnetic resonance (MR) signal (5%) restricted to
`the occipital cortex contralateral to the visual aura [94].
`These initial changes were followed by a decrease in the
`mean MR signal (by 5%), corresponding to the localised
`scotoma. The average velocity of
`the spread of
`the
`hypoperfusion over the cortex was 3.5 mm/min, being in
`concert with previous experimental studies [121]. In three
`spontaneous attacks of migraine with aura that were
`captured within 20 min of the onset of visual symptoms;
`the BOLD data revealed increases in the amplitude of the
`MR signal [94].Thus, this study supports previous reports of
`spreading depression as an initial cortical grey matter
`hyperaemia with a characteristic velocity that is followed
`by hypoperfusion. It lends support to studies in animals that
`the hypoperfusion spreads along the cortical surface at a
`relatively constant rate, sparing the cerebellum, the basal
`ganglia and the thalamus, and ultimately spanning the
`vascular distributions of the four major cerebral arteries
`[121]. A plausible explanation for the blood flow changes
`seen in association with the aura in a migraine attack is that
`they are the result of spreading depression—a transient
`marked reduction in electrical activity in the grey matter
`which advances across the cortical surface. The rate of
`advance is consistent with the spread of symptoms observed
`and is associated with decreases in blood flow [121].
`Spreading depression can move transcallosally to homolo-
`gous regions of the opposite hemisphere in animals, and
`transcallosal spread may account
`for
`the bilaterality
`observed at the onset of the headache [201]. One conclusion
`that has been raised from the studies is that the migraine
`aura is not evoked by ischemia, but evoked by aberrant
`firing of neurones and related cellular elements. An
`important question that can be raised is how the event is
`linked to activation of the trigeminovascular reflex [136].
`One tempting way would be to link the cortical spreading
`depression to neurogenic inflammation in the dura mater
`and from there activation of sensory and autonomic reflexes
`[18]. However, the dura mater is an extracerebral structure,
`separated from the brain by, e.g., CSF and is nourished by
`the external carotid artery [147]. Alternatively, specific cell
`bodies projecting from the brainstem to cerebral vessels
`such as the extensive adrenergic and serotonergic efferents
`from nuclei of the locus coeruleus and of the raphe nuclei,
`respectively, could be involved. In fact, there are some data
`to support
`this suggestion [19,46,161] showing close
`association between intracerebral nerve fibers and cerebral
`blood vessels. This has been examined at depth subse-
`
`quently revealing a direct neurogenic control by intrinsic
`serotonergic (5-HT) neurons on the cerebral microvascular
`bed [29].There exist close association between the 5-HT
`neurons and microarterioles, capillaries and perivascular
`astrocytes; this is more apparent in regions where manip-
`ulation of the intrinsic 5-HT neurons elicits uncoupling
`between flow and metabolism [28,29].
`In patients with migraine without aura, the situation is
`somewhat more intricate [199]. During attacks, small
`increases in blood flow were observed in the cingulate,
`auditory and visual association cortices, and in brainstem
`regions. These changes normalized after
`injection of
`sumatriptan and induced complete relief from headache as
`well as from phono- and photophobia. However,
`the
`changes were small and could only be significant if the
`PET data from all nine subjects were normalized, thus being
`by and large in agreement with previous negative studies
`with the xenon method which lacks the precision of PET
`[144]. Further support for the importance of a brainstem
`region was obtained in a patient that developed an attack of
`migraine without aura after glyceryl trinitrate administra-
`tion. Bahra et al. [13]observed activation in the dorsal rostral
`brainstem region and hence reproduced those data seen
`previously by Weiller et al. In addition, the authors observed
`a neuronally driven vasodilatation and activation of regions
`associated with pain processing [13,199].
`A PET study in patients with attacks of cluster headache
`and of capsaicin-induced head pain has reported blood flow
`changes that suggest, in part, a response that is primarily
`generated by the pain [132,133]. In this study, the anterior
`cingulate cortex was activated as would be expected, as part
`of the affective response. Activation was also seen in the
`frontal cortex, the insulae and the ventroposterior thalamus
`contralateral to the side of the pain. The only activated area
`that was particular to cluster headache was the ipsilateral
`hypothalamus. This region is important in the control of
`circadian rhythm and can be linked to the neurohormonal
`imbalance seen in cluster headache. This raised the possi-
`bility that the pathophysiology of cluster headache is driven
`partially or entirely from the CNS. The episodic nature of the
`disorder suggests involvement of at least the suprachiasmatic
`region, possibly associated with the human biological clocks.
`Spontaneous as well as nitroglycerin-induced cluster head-
`ache attacks were both associated with cerebral vasodilata-
`tion, interpreted as occurring via a neuronal mechanism
`[134]. Vasodilatation of the cranial vessels was not consid-
`ered to be specific to any particular headache syndrome, but
`generic to cranial neurovascular activation involving both
`sensory and parasympathetic reflex mechanisms as evi-
`denced previously by the release of the sensory neuro-
`transmitter CGRP and the parasympathetic messenger VIP in
`man [74] and experimentally in animal [75,204].
`PET scans of patients with acute attacks of cluster
`headache demonstrated an unilateral activation in the
`ipsilateral hypothalamic grey matter [132]. It is likely that
`the fundamental driving process arises in diencephalic
`
`3
`
`

`

`L. Edvinsson, R. Uddman / Brain Research Reviews 48 (2005) 438–456
`
`441
`
`pacemakers. While migraine and cluster headache share
`much in the expression of the pain, their underlying initiator
`mechanisms distinguish them.
`Indeed,
`it
`is the CNS
`triggering or driving process that ultimately characterizes
`many of the primary headache syndromes. In contrast, PET
`scans of capsaicin-induced pain or in migraine [133,199]
`showed no hypothalamic activation.
`In patients with
`capsaicin-induced pain blood flow changes were seen in
`an area consistent with the cavernous sinus/carotid artery
`just as there are blood flow changes in these vessels in
`cluster headache. This implies that the activation of the
`carotid artery does not relate specifically to cluster head-
`ache, but rather a trigeminovascular autonomic reflex. The
`flow changes may therefore be epiphenomena of
`the
`trigeminal activation, and not part of the disease generation
`process.
`A possible way to link recently documented alterations in
`the intracranial circulation to the genetic theory is via the
`observation that genetically defect ion channels may more
`easily be activated (due to altered membrane potential and/or
`function) and result in excitation of neurons in situations
`where they are exposed to excessive stress. Proof for
`involvement of brainstem nuclei in migraine came from a
`PET study by Weiller et al. [199] and has now been supported
`by others [13]. During acute attacks, increased local blood
`flow was observed in brainstem regions (specifically mid-
`brain and pons regions). The brainstem activation persisted
`after injection of sumatriptan. These findings support the idea
`that the pathogenesis of migraine (and the associated emesis)
`is related to an imbalance in the activity of brainstem nuclei
`regulating nociception and vascular control. On the other
`hand,
`it could equally well be an activation of the
`periaqueductal grey (PAG) acting as a filter to inhibit the
`pain [66]. The study revealed activation of the dorsal raphe
`nucleus (DRN) and the locus coeruleus (LC). It is well known
`that these centers have a dense supply of serotonergic and
`adrenergic fibers, respectively. The fibers may evoke vaso-
`constriction (via catecholamines or 5-HT) and hence explain
`the connection with the trigeminovascular reflex. Alterna-
`tively,
`the DRN and LC send descending fibers to the
`trigeminal nucleus caudalis (TNC) and dorsal root ganglia
`(DRG) where they act in a gate-control function and the PAG
`acts to inhibit this. Thus, sensory transmission associated
`with the TNC appears to be regulated by a complex system. It
`is still unclear whether the brainstem finding reveal the origin
`of the disease or if it is an accompanying activation designed
`to limit the symptoms of the migraine headache.
`
`3. The ion channel connection
`
`Clinical studies have revealed that migraine patients
`usually have a family history [84]. In the two main types of
`migraine, with aura and without aura, the familial aggrega-
`tion cannot be explained by simple mendelian inheritance
`patterns. FHM is the only variety of migraine in which a
`
`mendelian type of inheritance has been clearly established.
`A few years ago, a candidate region on chromosome 19 was
`identified as a gene that encodes an a1 A subunit of a
`voltage-gated P/Q-type calcium channel
`[110,111,150].
`FHM with cerebellar signs was subsequently linked to
`mutations in CACNA1A [15,35,69,150,188,195]. Thus, this
`type is now called FHM 1 and has been associated with
`mutations in CACNA1A [150], but in others, a second locus
`has been mapped on chromosome 1 [34,71] and is called
`FHM 2. In still other cases, the disorder is linked to neither
`site, suggesting the existence of a third locus, FHM 3 [34].
`Eight mutations in CACNA1A have been identified in 18
`families affected by hemiplegic migraine and in two patients
`with sporadic hemiplegic migraine [1,150]. CACNA1A is
`specifically transcribed in cerebellum, cerebral cortex,
`thalamus, hypothalamus and upper brainstem. The opening
`and closing of voltage-gated calcium channels are controlled
`by changes in voltage across the cell membrane and mediate
`the entry of calcium into the cell. These channels are of
`critical
`importance because the gradient between intra-
`cellular and extracellular calcium controls neurotransmitter
`release, neuronal excitation and other neuronal functions.
`The calcium channel
`is present in axons and dendrites,
`suggesting that it has both presynaptic and postsynaptic
`roles in modulating cell-to-cell communication.
`Several different missense mutations in the CACNA1A
`gene have been detected in unrelated FHM families [150].
`Generally, patients with FHM have missense mutations and
`these alter the gating properties of the channel [36,150]. The
`first FHM mutation (R192Q) occurs in a region that is
`believed to be part of the voltage sensor domain of the
`calcium channel. The second, most prevalent of the FHM
`mutations (T666M)
`is found within the pore-forming
`hairpin loop of the second domain of the channel. Two
`other FHM mutations (V714A and I1811L) are located in
`the transmembrane segments that may influence calcium
`channel inactivation, thereby blocking calcium transfer. The
`functional consequences of FHM 1 mutations are now
`receiving much attention subsequent to producing a knock-
`in mouse that carry the human FHM 1 R192Q mutation
`[196]. The researchers found gain-of-function effects that
`include increased CAR2.1 current density in cerebellar
`neurons, enhanced neurotransmission at the neuromuscular
`junction, and a reduced treshold and increased velocity of
`cortical spreading depression [196]. The data suggest that
`this mutation may result
`in increased susceptibility to
`cortical hyperexcitability and link spreading depression
`and aura in migraine.
`The gene for FHM 2 was recently identified when an
`Italian research group found two different missense muta-
`tions in the ATP A2 gene, coding for the alpha 2 subunit of
`the Na+,K+-ATPase in two families with pure FHM 2 [33],
`and this has been confirmed in two Dutch families [197].
`This subunit binds sodium, potassium and ATP, and utilizes
`ATP hydrolysis to extrude Na+ ions. The Na+ pumping
`provides the steep Na+ gradient essential for the transport of
`
`4
`
`

`

`442
`
`L. Edvinsson, R. Uddman / Brain Research Reviews 48 (2005) 438–456
`
`amino acids and calcium. Hypothetically, a mutation like
`this may result in loss of function which could make the
`brain more susceptible to spreading depression, however,
`more experimentation is needed.
`
`4. Nerves in the walls of intracranial vessels
`
`Since intracranial vessels are the only source for eliciting
`intracranial pain and in particular referred pain [159], the
`understanding of the vascular innervation by autonomic and
`sensory nerves is a prerequisite for the understanding of
`intracranial pain as it occurs in primary headaches. The
`intracranial blood vessels are supplied with nerve fibers that
`emanate from cell bodies in ganglia belonging to the
`sympathetic, parasympathetic and sensory nervous systems
`(Fig. 1) [92]. In addition, cerebral resistance vessels may be
`innervated by fibers that originate within the brain itself
`thereby representing an intrinsic nerve supply [40].
`
`4.1. Sympathetic nervous system
`
`The sympathetic nerves that supply the cerebral vessels
`arise mainly from the ipsilateral superior cervical ganglion
`[142], while some nerve fibers that supply the vertebral and
`basilar arteries originate from the inferior cervical ganglion
`and the stellate ganglion [3]. The activation of these fibers
`results in vasoconstriction, modulation of cerebrovascular
`autoregulation, reduction of intracranial pressure and a
`decrease of cerebral blood volume and cerebrospinal fluid
`
`production [40]. The responses are mainly mediated by
`noradrenaline (NA) and neuropeptide Y (NPY) [47,48], at
`least 40–50% of
`the NA-positive cells contain NPY
`[12,186].
`The neurotransmitter content in the nerve cell bodies is
`influenced by various factors: Activation may increase
`catecholamine synthesis and NPY mRNA [96], while
`denervation results in depletion of NA and NPY [48].
`However, some time after sympathectomy,
`there is an
`upregulation of NPY-containing fibers of parasympathetic
`origin [17]. Furthermore, there are age-dependent changes
`in sympathetic neurons; in old rats, there is a selective loss
`of NPY with a concomitant
`increase of nerve fibers
`containing vasoactive intestinal peptide (VIP) and calcitonin
`gene-related peptide (CGRP) around cerebral blood vessels
`[21]. In man there is a significant reduction with age of
`NPY, VIP, substance P (SP) and CGRP [52].
`Electronmicroscopic and functional studies have
`revealed that NA, NPY and adenosine triphosphate (ATP)
`are co-stored in large dense-cored vesicles [21]. Stimulation
`of the sympathetic nerves results in the release of these
`transmitters, the stimulus intensity determines the relative
`contribution of NA and NPY. At resting conditions, little
`NPY is released, and hence sympathetic vasoconstriction is
`largely due to adrenoceptor and purinergic receptors
`whereas in situations of high sympathetic activity,
`the
`contribution of NPY becomes prominent [129].
`It has been suggested that the small pial vessels on the
`cortical surface are supplied by NA-containing fibers
`emanating from an intracerebral source such as the LC
`
`Fig. 1. Schematic illustration of the perivascular nerves in intracranial arteries. Sympathetic nerves originate in the superior cervical ganglion and store
`noradrenaline, ATP and neuropeptide Y. The presynaptic fibres originate in the sympathetic chain. Parasympathetic nerves have their major origin in otic and
`sphenopalatine ganglia and store VIP, PACAP, nitric oxide and acetylcholine. Sensory fibers originate mainly in the trigeminal ganglion and store CGRP,
`substance P, neurokinin A, PACAP and nitric oxide [92].
`
`5
`
`

`

`L. Edvinsson, R. Uddman / Brain Research Reviews 48 (2005) 438–456
`
`443
`
`and/or the hypothalamus [37,151]. Support for this hypoth-
`esis comes from studies showing that destruction of the LC
`induces a reduction in the number of noradrenergic nerve
`fibers in intracerebral vessels [40] and that central stim-
`ulation of NA neurons in the hypothalamus is associated
`with an increase in hypothalamic blood flow which is
`unaffected by superior cervical ganglionectomy or by the h-
`adrenoceptor antagonist propranolol [163]. It is tempting to
`involve such a pathway in coupling neuronal activity to
`local blood flow regulation [127].
`
`4.2. Parasympathetic nervous system
`
`The bclassicalQ transmitter in parasympathetic nerves is
`acetylcholine (ACh) and their cell bodies contain acetyl-
`cholinesterase (AChE). Cerebral blood vessels have
`perivascular nerves
`that display AChE activity
`[42,97,180], and are choline acetyltransferase (ChAT)
`positive [164,180]. At the ultrastructural level, varicosities
`that contain numerous small agranular vesicles (40–60 nm
`in diameter) and that
`remain after sympathectomy are
`generally presumed to represent cholinergic nerve termi-
`nals [42]. These varicosities frequently occur
`in close
`apposition to large dense-cored vesicles in the neuro-
`effector area, thus suggesting that parasympathetic nerves
`have the potential
`to interact with sympathetic nerve
`terminals near cerebrovascular smooth muscle [41,44]. In
`several species, ACh induces constriction of
`isolated
`cerebral arteries when deprived of the endothelium, while
`transmural nerve stimulation predominantly induces relax-
`ation in the same preparations [122]. The neurogenic
`vasodilatation in these preparations is not blocked by
`atropine and is thus non-cholinergic [122,123]. One
`possible explanation is that additional substances are
`released together with ACh to mediate dilatation
`[122,123,164]. Several neuromessengers, which induce
`cerebral neurogenic vasodilatation, have been suggested.
`Among these are VIP, pituitary adenylate cyclase activat-
`ing polypeptide (PACAP) and nitric oxide (NO), which all
`seem to mediate a major component of the vasodilator
`responses of
`isolated cerebral arteries and in vivo as
`demonstrated by cerebral blood flow measurements
`[83,107,192]. In fact, it has been suggested that NO might
`be the last
`link in cholinergic transmission. Another
`possibility would be that ACh mainly acts prejunctionally
`to inhibit neurotransmitter release from autonomic nerves
`[124]. The vast majority of parasympathetic nerve fibers to
`cerebral vessels originate in sphenopalatine and otic
`ganglia [54,178].
`VIP was the first neuropeptide demonstrated in perivas-
`cular nerve fibers around brain vessels [120]. Other peptides
`of the VIP family, such as PHI (peptide histidine isoleucine)
`[41], its human form PHM (peptide histidine methionine)
`are seen in nerve fibers that supply cerebral vessels
`[52,61,192]. The distribution of VIP-immunoreactive nerve
`fibers varies between species.
`In most species, VIP-
`
`containing nerves are most abundant in the circle of Willis
`and the major cerebral arteries. The density of the nerve
`plexus is highest in the carotid system and diminishes in
`caudal direction. In man, the VIP-immunoreactive nerve
`supply is sparse in both cerebral arteries or veins [56]. In the
`human sphenopalatine ganglion, there is a rich supply of
`cell bodies containing VIP, PACAP and NO [193]. In
`addition, human sphenopalatine ganglia express mRNA for
`NPY Y1 and VIP1 receptors.
`In some locations, AChE activity and VIP immunor-
`eactivity can be seen in the same vascular nerve fibers, and
`this has led to the suggestion that VIP and ACh coexist in
`parasympathetic nerve endings [90,128]. Furthermore,
`immunoelectron microscopic studies revealed that, in the
`human superficial temporal artery, VIP immunoreactivity is
`localized exclusively to large electron-dense secretory
`vesicles (diameter 70–100 nm) in nerve terminal varicosities
`which also contain numerous smaller sized agranular
`vesicles (diameter 40–60 nm) presumed to represent para-
`sympathetic cholinergic neurons [59]. It should be noted,
`however, that other studies on the cholinergic and VIPergic
`cerebrovascular innervation have demonstrated that ChAT
`and VIP immunoreactivities are co-localised in less than 5%
`of the fibers examined [138].
`Pituitary adenylate cyclase activating peptide is a vaso-
`active peptide that displays 68% homology to porcine VIP
`and is about 1000 times more potent than VIP in stimulating
`adenylate cyclase activity in cultured rat anterior pituitary
`cells [4]. PACAP immunoreactivity and PACAP mRNA
`have been found in the sphenopalatine and otic ganglia
`[193]. Perivascular nerve fibers that contain PACAP
`immunoreactivity can be seen in cerebral blood vessels
`and PACAP mediates dilatation [107,168,192]. The major-
`ity of the PACAP-immunoreactive nerve fibers constitute a
`subpopulation of fibers containing VIP/NOS immunoreac-
`tivity as verified by tracing, denervation and co-localization
`experiments [61].
`NO is a highly labile molecule and information on its
`cellular localization has largely been attained by immuno-
`cytochemistry for nitric oxide synthase (NOS). There is
`evidence that NO is not only a candidate for
`the
`endothelium-derived relaxing factor in the endothelium,
`but also acts as a neurotransmitter [20]. NO is a non-
`conventional transmitter, since it appears to be released by
`diffusion rather than exocytosis upon formation,
`is not
`stored in vesicles, and its action is not dependent on the
`presence of conventional membrane-associated receptors.
`There is a rich supply of NOS-immunoreactive nerve
`fibers around cranial blood vessels from several species,
`including man [14,20,61,83,143,181,189,202].
`In the
`human circle of Willis, NOS-containing fibers a