In this review, we illustrate the main findings concerning basic mechanisms underlying CSD as neurophysiologic substrate of aura and report results on the effects on CSD from available drugs and new therapeutic options.
Clinical, neurophysiological and neuroimaging evidence of a link between migraine aura and CSD
About a third of migraine patients complain of transient focal aura symptoms beginning from minutes to hours before headache, or occurring during either the headache phase or even in its absence [
18]. Usually, migraine aura consists of fully reversible visual, sensory and/or dysphasic symptoms [
19]. These aura symptoms are accompanied by fully reversible motor weakness in hemiplegic migraine. This is referred to as familial (i.e., FHM) subtype when the condition is present in at least one first- or second-degree relative, and a sporadic subtype in the absence of family history [
20].
Specific genetic subtypes of FHM have been identified. Mutations of three genes all encoding ion-channels or membrane ionic pumps were discovered from 1996 to 2005. They involve a neuronal Ca
2+ channel (CACNA1A, FHM1), a glial Na
+/K
+ pump (ATP1A2, FHM2) and a neuronal Na
+ channel (SCN1A, FHM3), respectively [
13]. However, these mutations have not been identified in the more common types of migraine with typical aura. These forms are considered polygenic, with an overall heritability nearing 50%, and should be regarded as the result of the interaction of genetic and environmental factors [
21].
Descriptions of migraine with aura (MwA) from the year 1870 onwards have reported a slow, gradual progression of aura symptoms. Specifically, in 1941, Lashley, from the meticulous chartering of his own auras, suggested that aura symptoms reflect a cortical process progressing with a speed of 3 mm/min across the primary visual cortex [
22].
CSD was first described by Aristides Leão in 1944 [
23,
24]. Studying experimental epilepsy for his PhD thesis, Leão came across a depression of electroencephalographic (EEG) activity moving through the rabbit cortex at a rate of 3–6 mm/min after electrical or mechanical stimulations. The negative wave was sometimes preceded by a small, brief positivity, and always followed by a positive overshoot of 3–5 min [
25]. He observed that the threshold of CSD varied among cortical areas also in pigeons and cats, and once triggered, it spread in all directions. During CSD, neither sensory stimulation nor direct cortical stimulation evoked potential waves. Ongoing experimental seizure discharge was also suppressed by CSD, even if sometimes tonic-clonic activity preceded or followed SD [
23]. Based on similar propagation of the two processes, Leão hypothesized an association between CSD and seizures [
1].
Using microscopy and photography of pial vessels to assess cortical circulation, the researcher was also able to see both arteries dilated “as scarlet as the arteries” and veins, as a consequence of CSD. This latter observation, for the first time, indicated that the cerebral blood flow increase exceeded the increase in oxygen demand. This topic has become a matter of interest for all investigators studying changes in cerebral circulation over the course of CSD [
24].
In the early 20th century, aura was considered a vascular process involving an initial vasoconstriction followed by a reactive vasodilation responsible for head pain [
8,
26]. Later observations by Olesen et al. modified this idea by demonstrating a spreading reduction in cerebral blood flow, called “spreading oligemia”, occurring in patients with MwA [
27]. This finding completely redefined the underlying pathogenesis of aura by attributing blood flow changes during aura to changes in neuronal activity [
28].
Single photon emission computerized tomography (SPECT) [
29] and perfusion-weighted magnetic resonance imaging (MRI) [
30] studies have further supported the hypothesis that “spreading oligemia” observed during aura is primarily due to changes in neuronal activity. Additionally, perfusion abnormalities have been suggested to be a response of the autoregulatory mechanisms to underlying neuronal depression. However, in most SPECT and hyper-emia and subsequent spreading hypo-perfusion, patients never experienced symptoms of typical visual auras [
7,
31].
The first-ever study investigating occipital cortex activation during visual stimulation with functional magnetic resonance (fMRI) by blood oxygenation level (BOLD)-dependent contrast imaging in MwA patients demonstrated that the onset of headache or visual change (only in 2 patients), or both, were preceded by a suppression of the initial activation. This suppression slowly propagated into contiguous occipital cortex at a rate ranging from 3 to 6 mm/min. and was accompanied by baseline contrast intensity increases, indicating that vasodilatation and tissue hyper-oxygenation are associated with the induction of headache [
32]. Later, in 2001, Hadjikhani at al. [
33], strongly suggested that electrophysiological events consistent with CSD are involved in triggering aura in the human visual cortex. Using high-field fMRI with near-continuous recording during visual aura, the authors identified specific BOLD events in the visual cortex that were strictly linked to the aura percept, in both space (retinotopy) and time. Throughout the progression of each aura, unique BOLD perturbations were found in the corresponding regions of the retinotopic visual cortex. Like the progression of the aura in the visual field, the BOLD perturbations progressed from the paracentral to more peripheral eccentricities, in only the hemisphere corresponding to the aura. The source of the aura-related BOLD changes were localized in the extrastriate visual cortex (area V3A) rather than in the striate cortex (V1). Strikingly, the spread rate of the BOLD perturbations across the flattened cortical gray matter was consistent with previous measures of CSD.
As for diffusion changes on MRI, these have been especially observed in cases of prolonged complex migraine aura, suggesting cytotoxic edema in the absence of ischemic lesions [
34‐
37]. To this regard, it is noteworthy the finding of a spreading of cortical edema with reversibly restricted water diffusion from the left occipital to the temporo-parietal cortex in a case of persistent visual migraine aura [
38]. In another case of migraine with prolonged aura, hyper-perfusion with vasogenic leakage was detected by diffusion-weighted MRI [
39]. A further patient experienced a series of MwA attacks accompanied by slight pleocytosis and gadolinium (Gd-DTPA) enhancement in proximity of the left middle cerebral artery. In this patient, the migraine attacks and Gd-DTPA enhancement were reversed by prophylactic treatment [
40].
Magneto-electroencephalography (MEG) allows the study of direct current (DC) neuromagnetic fields in spontaneous and visually induced migraine patients. Few MEG studies have been conducted with this approach in MwA patients due to technical difficulties. The most relevant study to date has shown multiple cortical areas activated in spontaneous and visually induced MwA patients, unlike activation limited to the primary visual cortex in control subjects. This finding supports the hypothesis that a CSD-like neuro-electric event arises spontaneously during migraine aura or can be visually triggered in widespread regions of the hyper-excitable occipital cortex [
41].
MEG has also been used to directly register changes in cortical oscillatory power during aura. Specifically, alpha band desynchronization has been demonstrated with this technique in both the left extra-striate and temporal cortex over the period of reported visual disturbances. These terminated abruptly on the disappearance of scintillations, whereas gamma frequency desynchronization in the left temporal lobe continued for 8 to 10 minutes following the reported end of aura [
42].
Neuronal mechanisms of CSD in experimental models
When evoked by local extracellular K
+ concentrations exceeding a critical threshold, CSD is associated with the disruption of membrane ionic gradients. Both massive K
+ (with extracellular concentration increases up to 60 mM) and glutamate effluxes are believed to depolarize adjacent neurons to facilitate spread [
43]. Results from studies on ion-selective microelectrodes have shown that an extracellular K
+ rise is accompanied by falls in extracellular Na
+ and Cl
- during CSD, whereas water leaves the extracellular space with significant changes in extracellular pH [
44,
45].
Specifically, in chick retina, SD induced an initial increase in intracellular pH, which was associated with elevated levels of ADP, P-Creatine, lactate and pyruvate. This was followed by an intermediary acid shift, increases in ATP values and decreases in ADP, a late alkaline rebound, a decrease in P-Creatine levels, and elevations in both ADP and lactate levels. These transient changes in intracellular pH occurring parallel to changes of energy metabolite levels during SD, may be expressions of rapidly modifying metabolic activities of neurons and glial cells. The first alkaline shift was attributed to glial cells, whereas the intermediary acid shift was attributed to neurons. No specific cells were thought to be responsible for the late alkaline shift, which could explain the refractoriness of the neurons in this phase [
46]. Accordingly, in rat cerebellum, an initial decrease in [H
+] (pH increase) followed by a increase in [H
+] (pH decrease) was observed during SD [
45]. Further results obtained from a model of CSD showed acidification and a marked depression in the cortical energy status at the wavefront of SD. Afterwards, a residual activation of glycolysis and an accumulation of cGMP persisted for minutes after relatively rapid restorations of high-energy phosphates and pHi [
47]. Recovery from this process occurs usually within a few minutes without any tissue damage [
43,
45].
A causative link between enhanced glutamate release and facilitation of CSD, induced by brief pulses of high K
+, has been reported in mouse models of CSD [
48‐
51]. In some of these models, CSD could not be recorded after perfusing cortical slices by Ca
2+ free medium or after blocking Ca
2+ channels [
50,
52‐
55]. Glutamate involvement in CSD is further supported by the finding of CSD blockage by N-methyl- D-aspartate receptor (NMDA-R) antagonists, but not by non-NMDA-R antagonists,
in vivo
[
56‐
59] or in hippocampal and neocortical slices of rats [
5,
55]. CSD has also been reported to be blocked by the NMDA-R antagonist 2-amino-5-phosphonovaleric acid, in slices of human neocortical tissue [
60]. Furthermore, data have demonstrated that NR2B-containing NMDA-R are key mediators of CSD, providing the theoretical basis for the usefulness of memantine and some NR2B-selective antagonists for the treatment of MwA and other CSD-related disorders, such as stroke or brain injury [
50,
61].
According to the above findings, CSD cannot be induced in brain slices of FHM1 KI mice if either P/Q-type Ca
2+ channels or NMDA receptors are blocked. Conversely, blocking N- or R-type Ca
2+ channels seems to have only small inhibitory effects on the CSD threshold and velocity of propagation. This suggests that Ca
2+ influx through presynaptic P/Q-type Ca
2+ channels with consequent release of glutamate from cortical cell synapses and activation of NMDA-R is required for initiation and propagation of the CSD [
62]. This is in contrast with results of
in vitro and
in vivo pharmacological studies where CSD was induced by perfusing cortical slices with a high K
+ solution (rather than with brief K
+ pulses or electrical stimulation). In these models, NMDA-R antagonists only slightly increased CSD threshold without affecting its velocity. Accordingly, blocking P/Q-type (or the N-type) Ca
2+ did not significantly affect the CSD threshold obtained from perfusing cortical slices with progressively increasing K
+ concentrations [
51,
63]. Interestingly, removal of extra-cellular Ca
2+ did not block CSD but reduced it to about half the rate of propagation [
64].
Different results have been obtained for multiple CSD models induced
in vivo by continuous K
+ microdialysis or topical application of KCl, where P/Q-type (Cav2.1), or N-type, Ca
2+ channel blockers and NMDA-R antagonists led to a strongly reduced frequency, amplitude and duration, but not a complete suppression, of CSD events [
50,
65,
66]. Furthermore, Ca
2+ channel blockers have not been reported to affect CSD induced by pinprick
in vivo
[
66]. Therefore, results seem to be strongly influenced by the model used. Currently, electrical stimulation and/or brief applications of high K
+ are considered to be the most appropriate CSD-inducing stimuli, rather than prolonged applications of high K
+, for the better understanding of “spontaneous” CSD mechanisms occurring in migraine aura [
62]. Specifically, these models best reveal that the excitatory synaptic transmission, involved in CSD initiation and propagation at the pyramidal cortical cells, predominantly depends on presynaptic P/Q-type Ca
2+ channels.
Earlier studies reported that the Na
+ channel blocker TTX was not able to consistently inhibit CSD [
67‐
69]. More recently, Na
+ channels have been shown to be involved in the initiation of CSD in hippocampal slices [
5]. Their contribution to CSD was confirmed by Tozzi et al. [
70] in rat neocortical slices by reducing CSD propagation after applying the voltage sensitive Na
+ channel blocker TTX. In another study, Na
+ ion channel blockage was also seen to inhibit relative cerebral blood flow (rCBF) changes occurring during CSD induced on both cats and rats. In the same model, voltage-dependent Ca
2+ channel blockers had little effect on either the initiation or propagation of CSD spread, as was the case for ATP-activated K
+ channel blockers also [
71].
It has been demonstrated that the activation of alpha-amino-3-hydroxy-5-methyl-4-isoxazole propionate (AMPA) receptors (AMPA-R) can suppress the actions of NMDA-R in the neocortex [
72]. Earlier findings, however, suggested that NMDA-R blockers, but not AMPA-R antagonists, were able to inhibit CSD in rats [
70,
72,
73]. Conversely, a recent study has demonstrated that both 50 μM AMPA, as well as 10 μM of the NMDA-R antagonist 2-amino-5-phosphono-pentanoic acid (2AP5), significantly reduce the number of CSD cycles. Additionally, the gamma-aminobutyric acid (GABA)-mimetic drug clomethiazole (100 mg/kg i.p.) did not significantly affect the number of CSD cycles [
74]. Being so, the suppression of NMDA-R actions in the neocortex by AMPA-R activation, may represent an intrinsic protective mechanism against CSD and could, thus, be a potential therapeutic strategy against CSD-related neurological conditions including migraine aura.
In line with the above finding, AMPA-R, as well as GABA(A) and GABA(B)-R agonists, have been shown to inhibit cerebral blood flow changes associated to mechanically-induced CSD in all rats and in a proportion of cats. Furthermore, non-responders showed altered speeds of propagation and times to induction [
75]. In contrast, in a recent investigation, reproducible CSD episodes, induced by high extracellular K
+ concentrations in rat neocortical slices, were inhibited by antagonists of NMDA-R, but not by AMPA-R [
70]. Methodological differences (CSD models, dosages of agonists, outcome measures) could explain discrepancy in the results of the different studies carried out on this topic.
Recent autoradiographic findings suggest that selective changes in several receptor-binding sites, in both cortical and subcortical regions, are related to the delayed excitatory phase after CSD. In fact, in neocortical tissues, local increases of ionotropic glutamate receptors NMDA, AMPA, and kainate receptor binding sites have been observed. In addition, receptor binding sites of GABA(A), muscarinic M1 and M2, adrenergic alpha(1) and alpha(2), and serotonergic 5-HT(2) receptors were seen increased in the hippocampus. CSD also up-regulated NMDA, AMPA, kainate, GABA(A), serotonergic 5-HT(2), adrenergic alpha(2) and dopaminergic D1 receptor binding sites in the striatum [
76]. Therefore, not only glutamatergic mechanisms, but also changes in monaminergic and cholinergic pathways seem to be involved in CSD.
Vascular changes associated to CSD in experimental models
CSD has been reported to be associated with changes in the caliber of surface cortical blood vessels.
Leão was the first to report arteriole dilatation accompanying electrophysiological changes in CSD of rabbits [
24], which was later confirmed in rats and cats [
77,
78]. A further study using laser Doppler flowmetry, focusing on tissue perfusion rather than arterial diameter, has suggested that CSD is associated with an initial increase in cortical blood flow, which is thought to correspond to arteriolar dilatation [
79]. Triggering CSD results in a sustained wave of reduced cortical blood flow after initial vasodilation, as shown by single modality blood flow measurements, including autoradiographic methods [
80,
81] and laser Doppler flowmetry [
82]. Moreover, sustained hypo-perfusion was accompanied by a concurrent reduction in reactivity to vasoactive stimuli [
83]. Dual modality methods, such as laser Doppler flowmetry and extracellular electrophysiology, allowed for the concurrent assessment of changes in neuronal firing and cerebral blood flow in CSD but lacked parallel spatial and temporal resolutions [
84]. Optical intrinsic signal (OIS) imaging also enables visualization of CSD on the cortical surface with high temporal and spatial resolution [
85‐
87]. The optical correlates of CSD have been evaluated on both a mouse and a rat model by Ayata et al. [
88]. Vascular response to CSD propagates with temporal and spatial characteristics, which are distinct from those of the underlying parenchyma, suggesting a distinct mechanism for vascular conduction.
Using OIS imaging and electrophysiology to simultaneously examine the vascular and parenchymal changes occurring with CSD in anesthetized mice and rats, Brennan et al. [
89] observed vasomotor changes in the cortex which travelled at significantly greater velocities compared to neuronal changes. This observation further reinforces the idea that dissociation between vasomotor and neuronal changes during CSD exists. Specifically, dilatation travelled in a circuitous pattern along individual arterioles, indicating specific vascular conduction as opposed to concentric propagation of the parenchymal signal. This should lead to a complete rethinking of flow-metabolism coupling in the course of CSD. Vascular/metabolic uncoupling with CSD has also been reported by Chang et al. using a combination of OIS imaging, electrophysiology, K
+-sensitive electrodes and spectroscopy in mice [
90]. The authors identified two distinct phases of altered neurovascular function. In the first phase of the propagating CSD wave, the DC shift was accompanied by marked arterial constriction and desaturation of cortical hemoglobin. After recovery from the initial CSD depression wave, a second phase was identified where a novel DC shift appeared to be accompanied by arterial constriction and a decrease in tissue oxygen supply, lasting at least an hour. Persistent disruption of neurovascular coupling was supported by a loss of consistency between electrophysiological activity and perfusion.
Nitric oxide (NO) may play a relevant role in determining changes in cerebro-vascular regulation following CSD. In fact, the NO precursor L-arginine prevented the development of prolonged oligemia after CSD but had no influence on a marked rise of CBF during CSD. Moreover, rats treated with L-arginine recovered their vascular reactivity to hyper-capnia after CSD much faster than controls [
91]. The NO donor, 2-(N,N-diethylamino)-diazenolate-2-oxide (DEA/NO) had little effect on CSD but reversed the effects of NO synthase (NOS) inhibition by 1 mM L-NAME, in a concentration-dependent manner, suggesting that the increased formation of endogenous NO associated with CSD is critical for subsequent, rapid recovery of cellular ionic homeostasis. Molecular targets for NO may be either brain cells, through the suppression of mechanisms directly involved in CSD or local blood vessels by means of coupling flow with the increased energy demand associated with CSD.
The potent vasoconstrictor endothelin-1 (ET-1) applied on rat neocortices has been demonstrated to induce CSDs through the ET(A) receptor and phospholipase C (PLC) activation. Primary targets of ET-1 mediating CSD seem to be either neurons or vascular smooth muscle cells [
92]. This finding provides a bridge between the vascular and the neuronal theories of migraine aura. However, the micro area of selective neuronal necrosis, induced by ET-1 application suggests a role by vasoconstriction/ischemia mechanisms. This observation contrasts with the lack of neuronal damage in several CSD models [
93].
Genetic evidence of CSD involvement in migraine
Genetic factors are known to enhance susceptibility to CSD, as results from transgenic mice expressing mutations associated with FHM or cerebral autosomal dominant arteriopathy with subcortical infarcts and leuko-encephalopathy (CADASIL) have shown [
94‐
99]. Specifically, P/Q-type Ca
2+ channels, located in somato-dendritic membranes and presynaptic terminals in the brain, play a pivotal role in inducing potential-evoked neurotransmitter release at CNS synapses [
100]. Missense mutations in the gene encoding the pore-forming α1 subunit of voltage-gated P/Q-type Ca
2+ channel, responsible for the rare autosomal dominant subtype of MwA FHM1, induce a gain-of-function of human recombinant P/Q-type Ca
2+ channels, due to a shift to channel activation at lower voltages [
101]. Increased P/Q-type Ca
2+ current density in cortical pyramidal cells has been demonstrated in Knock-in (KI) mice carrying FHM1 mutations [
101‐
103]. Furthermore, FHM1 KI mice have shown a reduced threshold for CSD induction and an increased velocity of CSD propagation [
63,
104]. These mice represent a powerful tool for exploring presynaptic regulation associated with expression of P/Q-type Ca
2+ channels. Mutated P/Q-type Ca
2+ channels activate at more hyper-polarizing potentials and lead to a gain-of-function in synaptic transmission. This gain-of-function might be responsible for alterations in the excitatory/inhibitory balance of synaptic transmission, favoring a persistent state of hyper-excitability in cortical neurons which may increase the susceptibility for CSD [
101]. In contrast, spontaneous CACNA1a mouse carrying mutations producing partial loss-of-function of the P/Q-type Ca
2+ channel, need approximately a 10 fold higher electrical stimulation intensity in order to evoke a CSD compared to wild-type mice [
105].
FHM2, the autosomal dominant form of MwA, is caused by mutations of the α2-subunit of the Na
+,K
+-ATPase, an isoform almost exclusively expressed in astrocytes in the adult brain. In a FHM2 KI mouse model carrying the human W887R mutation in the Atp1a2 orthologous gene,
in vivo analysis of CSD in heterozygous F Atp1a2 (+/R887) mutants revealed a decreased induction threshold and an increased velocity of propagation. While several lines of evidence suggest a specific role on the part of glial α2 Na
+/K
+ pump in active reuptake of glutamate from the synaptic cleft, it is plausible that CSD facilitation in the FHM2 mouse model is sustained by inefficient glutamate clearance by astrocytes, leading to an increase in cortical excitatory neurotransmission [
106].
MwA is often the first manifestation of cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL), caused by NOTCH3 gene mutations expressed predominantly in vascular smooth muscles. In a recent study, CSD was reported to be enhanced in mice expressing either a vascular Notch 3 CADASIL mutation (R90C) or a Notch 3 knock-out mutation. These findings further support the role of the trigeminal neurovascular unit in the development of migraine aura [
107].
Astrocytes and gap-junction involvement in CSD
Astrocytes, a subset of glial cells, reside next to neurons, establishing together a highly interactive network [
121]. Astrocytes play a pivotal role in limiting CSD by acting as a buffer for the ionic and neurochemical changes which initiate and propagate CSD [
122]. On the other hand, astrocyte interconnections are believed to contribute to propagating the CSD wave, by way of K
+ liberation, allowed for by an opening of remote K
+ channels. Moreover, energy failure in astrocytes increases the vulnerability of neurons to CSD [
123]. There is increasing evidence suggesting that, while synapses connect neuronal networks, gap-junctions most likely connect astrocyte networks [
124]. Clusters of these tightly packed intercellular channels allow for the direct biochemical and electrical communications among astrocytes, contributing to a syncytium-like organization of these cells [
125]. Membranes of adjacent astrocytes have connexin-containing hemi-channels which can bridge an intercellular gap to form a gap-junction [
126]. This interaction between the two hemi-channels opens them both, allowing for the intercellular passage of ions and small molecules [
127]. Approximately 1.0–1.5 nm in diameter, gap-junctions permit the transport of molecules up to about 1 kDa in size. Astrocytes express at least three different connexins at gap-junctions with regional differences in their distributions.
Experimental studies have suggested an involvement of gap-junctions in CSD by regulating the
milieu around active neurons including extracellular K
+, pH and neurotransmitter levels (especially glutamate and GABA), as well as propagating intercellular Ca
2+ waves [
128]. Non-junctional connexin hemi-channels may also contribute to the release of adenosine triphosphate (ATP). This extracellular messenger is able to mediate Ca
2+ wave propagation directly or via the transfer of a messenger which triggers ATP release from one cell to another [
129]. Generation and propagation of CSD may depend on neuronal activation and Ca
2+ influx triggered by NMDA-R. Interestingly, NMDA-R antagonists block CSD but, unlike the gap-junction blockers, do not inhibit Ca
2+ wave propagation.
Astrocytes are known to express several types of glutamate receptors, including NMDA-R. Glutamate release from astrocytes has also have been reported to be mediated via the opening of connexin hemi-channels [
127]. For this, gap-junction-mediated propagation of Ca
2+ waves may represent the advancing front of CSD, contributing to the triggering of the secondary depolarization of the surrounding neurons, leading to further releases of K
+ and glutamate into the extracellular space. Glutamate may then stimulate cytosolic Ca
2+ oscillations in astrocytes, providing a feedback loop involved in CSD propagation. If so, gap-junction blockage would represent a viable pharmacological strategy for MwA prevention. Evidence of a gap-junction coupling Ca
2+ waves between pia-arachnoid cells and astrocytes has also been reported, suggesting a transfer of Ca
2+ signals from cells of the cortical parenchyma into the meningeal trigeminal afferents, all of which might mediate the induction of neurovascular changes responsible for migraine headache [
130].
The relationship between CSD and headache
Recent electrophysiological data has provided direct evidence that CSD is a powerful endogenous process which can lead to persistent activation of nociceptors innervating the meninges. Regardless of the method of cortical stimulation, CSD in rat visual cortices induces a two-fold increase in meningeal nociceptor firing rates, persisting for 30 min or more. Meningeal nociceptors represent the first-order neurons of the trigemino-vascular system, whose activation is involved in the initiation of migraine headache [
137]. CSD waves moving slowly across the cortex can promote the releases of K
+, arachidonic acid, hydrogen ions, NO and ATP. Critical levels of these substances are thought to cause sensitization and activation of trigeminal neurons in the afferent loop and, in turn, activate second-order neurons in the trigemino-cervical complex. These second-order neurons transmit sensory signals to the brainstem and parasympathetic efferents, the latter projecting from the sphenopalatine ganglion. CSD has been suggested to promote persistent sensitization, thereby provoking the activation of meningeal nociceptors through a mechanism involving local neurogenic inflammation, with contribution of mast cells, macrophages and the release of inflammatory mediators. Local action of such nociceptive mediators increases the responsiveness of meningeal nociceptors. Recent research has provided key experimental data suggesting the role of complex meningeal immuno-vascular interactions leading to an enhancement in meningeal nociceptor responses [
137]. CSD also induces increased neuronal activity of central trigemino-vascular neurons in the spinal trigeminal nucleus (C1-2) as measured by single-unit recording. It therefore represents a "nociceptive stimulus" capable of activating both peripheral and central trigemino-vascular neurons underlying the headache phase of MwA [
137].
Recent evidence suggests that central trigeminal neurons are activated by CSD. Specifically, an increase in the spontaneous discharge rate, following the induction of CSD by cortical injection of KCl was not reversed through the injection of lignocaine into the trigeminal ganglion 20 min after CSD induction. Lignocaine injection prior to the initiation of CSD also failed to prevent the subsequent development of CSD-induced increases in discharge rates [
138]. In these experiments, lignocaine at a dosage of 10 μg (capable of interrupting stimulus-induced responses to either electrical stimulation of the
dura mater or mechanical stimulation of the craniofacial skin) reduced basal the discharge rate of second-order trigeminovascular neurons. This increased traffic in the second-order neurons induced by CSD, however, was not influenced by the blockage of conduction in first-order neurons which was due to lignocaine injection into trigeminal ganglion after CSD induction by cortical pinprick. A time point of 20 min post-lignocaine injection was chosen because responses to evoked stimulation reached a minimum at this time.
It has been suggested that CSD may produce a rapid sensitization at first sensory neurons which could become “locked-in” and, therefore, would not be influenced by a later reduction in sensory traffic, like that induced by the injection of lignocaine into the trigeminal ganglion [
139]. An increase in discharge rate produced by CSD has also been observed when lignocaine is injected into trigeminal ganglion, prior to the induction of CSD. This is further evidence that CSD does not act solely by increasing continuous traffic in primary trigemino-vascular fibers through a peripheral action alone, but rather exerts its effect through a mechanism intrinsic to the CNS. Accordingly, pain in MwA may not always be the result of peripheral sensory stimulation, but may arise via a central mechanism [
140].
The principal opposition to this hypothesis is based upon the belief that mediators released as a consequence of CSD induction cannot be sustained in the perivascular space to induce persistent trigeminal sensitization and the subsequent hours-lasting headache because of the
glia limitans barrier (astrocyte foot processes associated with the parenchymal basal lamina surrounding the brain and spinal cord, regulating the movement of small molecules and cells into the brain parenchyma) and the continuous cerebrospinal fluid (CSF) flow [
141]. Additionally, the delay of 20–30 min between aura and headache suggests that a time lag is required for the transduction of algesic signals beyond
glia limitans via inflammatory mediators. In support to this rebuttal, Karatas et al. demonstrated that intense depolarization and NMDA receptor overactivation due to CSD, opens neuronal Pannexin1 (Panx1) mega-channels [
142]. Panx1 activation induces a downstream inflammasome formation involving caspase-1 activation and the sustained release of pro-inflammatory mediators from
glia limitans such as high-mobility group box 1 (HMGB1) and IL-1β, both of which take part in the initiation of the inflammatory response [
143‐
146]. A subsequent NF-kB translocation was observed inside the cortex, involving astrocytes, forming or abutting
glia limitans, followed by the activations of both cyclooxygenase (COX)2 and inducible NOS (iNOS). The inhibition of Panx1 channels or HMGB1 resulted in a reversal of this effect.
A CSD-induced neuronal megachannel opening may therefore promote sustained stimulus required for both sensitization and activation of meningeal trigeminal afferents through the maintenance of inflammatory responses which may be involved in the subsequent headache pain.