Introduction

There is growing evidence that cortical spreading depression (CSD) underlies migraine aura [1] and possibly acts as a trigger for headache as well [2, 3]. CSD is a slowly propagating depolarization of neuronal and glial membranes evoked when extracellular K+ concentrations ([K+]e) increase above a genetically determined threshold, for example, by transient ischemia [4]. The threshold for initiation of CSD presumably is lower in migraineurs than in the normal population and possibly is linked to overall cerebral hyperexcitability. Indeed, interictal cortical excitability is increased in migraineurs [5], and there is a clinical association between migraine and epilepsy [6]. The genetically determined CSD threshold is modulated further by endogenous (eg, gonadal hormones) and exogenous factors (eg, migraine prophylactic drugs) [7•, 8, 9, 10••]. Hence, the susceptibility of brain tissue to CSD (and thus migraine) is governed by complex interactions among multiple independent triggers, risk factors, and modulators, some of which are reviewed in the following text.

Basic Features of CSD

CSD, originally described by Leão [11], is an intense depolarization of neuronal and glial membranes due to a sudden loss of membrane resistance and ionic gradients. It is characterized by cessation of all spontaneous or evoked synaptic activity and massive K+ efflux causing [K+]e to rise above 40–50 mM, exacerbated by cell swelling and shrinkage in extracellular space as a result of water entering the cells coupled to Na+ influx. High [K+]e in the tissue undergoing CSD is believed to depolarize adjacent brain tissue, and in this way, the CSD wave propagates into contiguous gray matter at an average speed of 3 mm/min, regardless of functional divisions or arterial territories [12]. Large Ca2+ influx and glutamate efflux add momentum to the spreading wave of depolarization [13].

Although little is known about the triggers of migraine attacks in humans, CSD is initiated when local [K+]e exceeds a threshold of approximately 10–12 mM in a minimum critical volume of brain tissue (∼1 mm3 in rodents). This can be achieved by cerebral ischemia and trauma or direct cortical application of excitatory amino acids, Na+/K+-pump inhibitors, or concentrated K+ solutions in both lissencephalic [11] and gyrencephalic animals [14], as well as in humans [15, 16, 17•]. Triggering CSD often requires activation of the N-methyl-d-aspartate (NMDA) subtype of glutamate receptors [18, 19], which, as noted earlier, adds momentum to the depolarization wave facilitating its spread. Direct intercellular transfer of ions and small molecules through gap junctions also may facilitate the spread.

CSD is associated with characteristic cerebral blood flow (CBF) changes: an initial, species-dependent reduction in CBF is followed by hyperemia, reaching up to 200% of baseline in some species, and then by oligemia (40–90% of baseline), which usually lasts up to an hour. During this post-CSD oligemic phase, neurovascular coupling and hypercapnic hyperemia are impaired [20]. Both the CSD and the post-CSD oligemia may cause relative tissue hypoxia [21]. Beyond the vasomotor effects, CSD affects neurotransmitter and neuromodulatory systems, second messenger cascades, immediate early genes, and growth factors, as well as inflammatory mediators such as interleukin-1β and tumor necrosis factor-α [22, 23].

Despite extensive investigation of CSD in experimental animals in vivo and in vitro [11, 12, 24, 25] and in human neocortical and hippocampal tissues in vitro [26, 27], evidence supporting CSD in human brain in situ was obtained only recently, using functional MRI during migraine [28] and epidural and intracortical recordings in injured brain [29–32].

CSD and Aura

Leão [33] and Milner [34] both pointed out the similarity between the velocity of CSD propagation and the march of visual aura reported by Lashley [35]. As with CSD in mammalian cortex, the speed of spread of migraine aura is approximately 3 mm/min. Aura is characterized by negative symptoms (eg, loss of sensation, visual scotoma) often preceded by positive symptoms (eg, paresthesias, visual scintillations), reminiscent of the cessation of cortical neuronal activity preceded by a brief burst of action potentials at the CSD wavefront. The first compelling evidence for the CSD theory of migraine in human brain was obtained using intra-arterial 133Xe tomography during aura. A slowly spreading oligemia (25–35% below baseline) was found propagating anteriorly from the occipital cortex with a speed similar to that of CSD [36, 37]. In some cases, the oligemia was preceded by focal hyperemia resembling the hemodynamic effects of experimental CSD summarized earlier. The temporal association between cerebral hemodynamic changes and migraine symptoms suggested that oligemia was probably not the cause, but rather a consequence of a primary upstream event, presumably CSD. These findings subsequently were corroborated by positron emission tomography [38] and magnetoencephalography [39]. However, the strongest piece of evidence emerged from high-field strength functional MRI (fMRI) studies during aura. Brain oxygen level-dependent (BOLD) imaging, which detects the ratio of non-paramagnetic oxygenated hemoglobin to paramagnetic deoxyhemoglobin as a rough measure of the difference between oxygen delivery and consumption, forms the basis for fMRI [40, 41]. A focal increase in BOLD signal was detected during visual aura, spreading within occipital cortex at a rate of 3.5 mm/min in a fashion that was retinotopically congruent with the patient’s aura symptoms. This initial BOLD increase was followed minutes later by a decrease, suggesting a rise and then a fall in CBF [28]. These and other lines of evidence [8] strongly suggest that CSD is the electrophysiologic substrate of migraine aura.

Evidence for an Enhanced Cortical Excitabilityin Migraineurs

It is not clear how CSD is triggered in migraineurs. However, because most factors that facilitate CSD occurrence are excitatory, depolarizing events (eg, NMDA receptor activation, intense neuronal activity such as seizures), it is plausible to hypothesize that migraineurs have enhanced cerebral excitability. In support of this hypothesis, migraineurs exhibit a reduced threshold to evoked phosphenes (bright scintillations) after transcranial magnetic stimulation. Patients with probable chronic migraine reportedly show increased transcranial magnetic stimulation-induced excitability, which may correlate with the frequency of their migraine attacks [42].

Mutations in genes encoding ion channels and pumps expressed by neurons or glia have implicated glutamatergic neurotransmission as a critical modulator of CSD threshold and cortical hyperexcitability in the pathophysiology of migraine. Familial hemiplegic migraine (FHM) is an autosomal dominant migraine syndrome characterized by severe and prolonged auras (eg, unilateral motor deficits, coma); more than 60% of FHM patients also suffer from migraine attacks with typical aura or without aura, making FHM a useful genetic model to explore the mechanisms of more common types of migraine as well. Numerous FHM mutations in three different genes have been identified, all of which appear to enhance cerebral excitability via different mechanisms (Table 1).

Table 1 Familial hemiplegic migraine syndromes
Full size table

FHM type 1 is caused by gain-of-function mutations in the CACNA1A gene on chromosome 19p13 encoding the α1 pore-forming subunit of Cav2.1 voltage-gated Ca2+ channels (P/Q type) expressed on both presynaptic terminals and somatodendritic membranes. Mutant channels open at lower depolarizing voltages so that Ca2+ influx is enhanced through single Cav2.1 channels under depolarizing conditions, and neurotransmitter release presumably is increased [43]. The control of transmitter release by Cav2.1 channels is much more prevalent in excitatory (eg, glutamate) than inhibitory (eg, γ-aminobutyric acid) synapses [44, 45], and an enhanced probability of glutamate release has been demonstrated in cortical pyramidal cell synapses of FHM1 mutant mice [46••]. Therefore, FHM type 1 mutations are believed to enhance cerebral excitability by both increasing glutamate release and amplifying the postsynaptic depolarization [46••, 47, 48]. FHM type 1 accounts for 50% of all FHM families [49].

FHM type 2 mutations in the ATP1A2 gene on chromosome 1q23 also might increase ambient [K+]e and glutamate in the synaptic cleft as a result of a loss of function of the α2 Na+/K+ adenosine triphosphatase (ATPase), a P-type ion pump that uses adenosine triphosphate to actively countertransport Na+ and K+ ions. FHM type 2 point mutations produce substitutions of conserved amino acids in important functional regions, such as the intracellular four to five loop containing the nucleotide-binding domain and the extracellular seven to eight loop, which is responsible for β subunit binding [50]. The Na+/K+ ATPase generates the ion gradients that maintain resting membrane potential and cell volume, regulate resting and activity-induced [K+]e levels, and provide the driving force for nutrient and neurotransmitter uptake [51].

The FHM type 3 missense mutation (Gln1489Lys) in the SCN1A gene on chromosome 2q24 encoding the α1 subunit of NaV1.1 voltage-gated Na+ channels [52] accelerates channel recovery from fast inactivation, an effect predicted to increase dendritic excitability and neuronal firing rates [53]. As a consequence, a relatively weak depolarizing stimulus (perhaps without consequences in healthy individuals) may cause excessive neuronal firing that increases [K+]e above the critical value that triggers CSD. Therefore, elevated extracellular glutamate and [K+]e may be a common mechanism of reduced CSD threshold in migraine [54].

CSD Susceptibility as a Consequence of Enhanced Excitability in Migraine

Experiments on mutant mouse models provide mechanistic insight into modulation of CSD susceptibility as a relevant mechanism in migraine. Spontaneously occurring or engineered mutations within the CACNA1A gene modulate CSD threshold. For example, the tottering mutation (a proline-to-leucine substitution in the S5-S6 linker region of repeat domain II of the α1 subunit of the P/Q-type Ca2+ channel) causes loss-of-channel function, impaired presynaptic Ca2+ influx, and reduced neurotransmitter release, mainly inhibiting excitatory neurotransmission [45, 55–58]. Tottering mice show a 10-fold resistance to CSD, with a slower CSD propagation speed and a failure to sustain regenerative spread of the depolarization wave [44]. In contrast, mutant mice expressing human FHM type 1 mutations in the same gene (R192Q or S218L) show increased CSD susceptibility [7•, 59, 60]. In accordance with the more severe clinical phenotype associated with the S218L mutation, CSD susceptibility was even higher in S218L mutant mice than in R192Q mutants, and S218L mutant mice developed coma and seizures in addition to the transient hemiplegia observed in both mutants in response to CSD [7•]. Furthermore, FHM type 1 mutant mice showed a facilitated subcortical propagation of CSD [7•], providing a potential explanation for the hemiplegia, seizures, and coma in FHM patients [61, 62].

Consistent with the female preponderance of nonhemiplegic migraine with aura and familial as well as sporadic hemiplegic migraine [63, 64], female mice expressing the R192Q or S218L mutation exhibited a significantly higher susceptibility than male mice toward CSD [7•]; this sex-related difference was completely abrogated by ovariectomy and “postmenopausal” age. Estrogen augments the effect of glutamate on Purkinje cells and the sensorimotor system [65], and estrogen and progesterone reportedly evoke changes in excitability during transcranial magnetic stimulation [66]. Interestingly, androgens appear to influence CSD in the opposite direction in FHM type 1 mutant mice: castration increased KCl-induced CSD frequency and propagation speed, and this was prevented by testosterone replacement in an androgen receptor-dependent manner [9]. Although the mechanisms linking sex steroids to migraine remain poorly understood, these data suggest that genetic and hormonal factors modulate both CSD and migraine susceptibility (Fig. 1).

Fig. 1
figure 1

A simplified scheme for the cortical spreading depression (CSD) theory of migraine. In susceptible brains, diverse triggers, such as intravascular microemboli and head trauma, can evoke CSD and lead to migraine aura and headache. The overall CSD susceptibility—that is, the likelihood of triggering CSD upon a given stimulus type and intensity—is enhanced (red) or suppressed (blue) by endogenous factors, such as monogenic or polygenic determinants and sex hormones, and exogenous factors known to influence migraine susceptibility, such as weather, fasting, specific types of food, sleep deprivation, stress, and drugs. Thus far, only the last item (ie, migraine prophylaxis) has been shown to affect CSD susceptibility; the others remain to be tested. FHM—familial hemiplegic migraine

Full size image

Microembolism as a Possible Endogenous Trigger for CSD

Migraine is associated with an increased risk of stroke, especially in younger women suffering from migraine with aura [67–69]. Patent foramen ovale is more common in migraineurs with aura, and vice versa [70]. Cardiac or pulmonary arterial defects (eg, patent foramen ovale, pulmonary arteriovenous malformations) with right-to-left shunts that increase the risk of cryptogenic stroke also are associated with a higher incidence of migraine [71]. Closure of such defects may reduce or abolish migraine attacks [72]. These findings imply that subclinical cerebral ischemia due to paradoxic embolism may trigger a migraine attack [73]. Indeed, cerebral ischemia is a well-known experimental trigger of CSD. Recent experimental data clearly demonstrate that cerebral microembolism can initiate CSD without causing lasting tissue injury [4]. Therefore, in susceptible brains, CSD may be triggered by transient mild hypoperfusion events, and less frequently, such events may lead to microinfarcts, placing migraine and stroke on a spectrum of vascular complications. If true in humans, vascular and hematologic disorders may act as migraine triggers.

CSD as a Headache Trigger

Besides causing the aura, CSD also has been proposed as a trigger of headache [74]. CSD can activate the meningeal trigeminovascular system and downstream pain pathways in rodents and lead to meningeal inflammation and plasma extravasation [2, 3, 75]. Although the mechanism of activation remains unknown, it has been assumed that H+, K+, NO, and other agents released into the extracellular space during CSD depolarize or otherwise activate adjacent perivascular trigeminal nerve endings surrounding local blood vessels. These changes may be facilitated by CSD-induced activation of matrix metalloproteinases and mild disruption of the blood–brain barrier [76]. CSD as a headache trigger is supported further by data showing that migraine prophylactic drugs, which are efficacious in preventing migraine attacks with or without a perceived aura, decreased CSD susceptibility [10••]. Reminiscent of the gradual buildup of their clinical efficacy over weeks to months, these drugs required chronic treatment to suppress CSD in experimental animals and did not show efficacy when tested acutely after a single dose.

Conclusions

A large body of clinical and experimental evidence implicates CSD as the electrophysiologic event underlying migraine aura and possibly the headache. Enhanced excitability appears to be a common theme in brains susceptible to migraine, which may translate into enhanced CSD susceptibility. Animal studies point toward genetic and hormonal factors modulating CSD susceptibility and suggest that transient mild ischemic events may trigger CSD without residual tissue damage. Improved understanding of the clinically relevant triggers and modulators of CSD will help researchers design targeted therapies for migraine.