Aura and Head pain: relationship and gaps in the translational models

Cortical spreading depression (CSD) is a pathophysiological phenomenon of gray matter characterized by sustained electrical silence and suppressed activity propagating in the cerebral cortex preceded by a brief shower of population spikes [1‐3]. CSD has been shown to occur in traumatic brain injury, ischemic stroke and subarachnoid hemorrhage in the human cerebral cortex [4‐8]. CSD has also been recognized as the biological substrate of migraine aura [9‐11]. Experimental animal studies have provided mechanisms that possibly link CSD to the activation of trigeminal neurons mediating lateralized head pain [12]. The probable mechanism between CSD and lateralized head pain in anesthetized rats is over-extrapolated to migraine without aura and a ‘silent aura’ theory is proposed for migraine without aura to explain headaches. Some basic CSD features unfortunately do not match many clinical features of migraine headache well. Those disregarded grey areas are briefly discussed here in order to understand the exact role of CSD in migraine and to develop a new perspective into the mechanisms of the disease.

CSD is demarcated by a slowly propagating wave of depressed electrical activity in the affected cerebral cortical region [1, 2, 13]. The ‘depression’ or ‘silence’ is actually as a result of a state of sustained depolarization where the membrane is captured/clamped at the depolarized potential with loss of excitability for a few minutes. The electrical silence dominates in the cortical region as a prolonged membrane depolarization and prevents action potentials and synaptic transmission. Sudden collapse of ionic gradients across membranes is associated with massive movements of Na+, Cl-, Ca++ and water into the cell, K+ accumulation outside the cell, pH decrease to near ischemic levels and decrease of extracellular space by 50% [2, 13]. Such a toxic depolarization state of the tissue in CSD has nothing in common with depolarization of a neuron generating AP, except for the word of ‘depolarization’ itself (Table 1) [1‐3, 13]. CSD happens in larger scale, lasts ~ 240,000 ms, causes up to 30 mV shift in the extracellular potential, covers every cell, and is characterized by loss of excitability of membranes in the tissue, while AP happens in micro scale, lasts 1 ms, causes no change in the extracellular potential, and is characterized by induced excitability state (Table 1) [1‐3, 13]. The CSD process leads to an enormous release of intracellular contents, neurotransmitters, molecules, vasoactive and nociceptive substances including glutamate, GABA, aspartate, glycine, adenosine, catecholamines to the extracellular space in the depolarized cortical region. Such a massive movement of ions, water and substances across membranes change the extracellular potential and is reflected electrophysiologically as a huge transient shift up to 30 mV in direct current (DC shift) [1, 2]. Conventional alternating current coupled devices cannot detect these slow discharges unless the filter is changed to allow very low frequencies.

The depolarization of CSD is reversible if the tissue is under physiological conditions and not associated with compromised energy/ blood flow. The contiguous spreading nature of DC shift comes from the fact that loss of membrane resistance and release of neurotransmitters, K+ and other molecules during depolarization induce a domino effect on adjacent cells. The improvement of electrical depression begins within a few minutes, though full recovery of spontaneous synaptic activity takes hours [2, 13].

The CSD induced depolarization state is not associated with increased excitability where action potentials fire, such as in epileptic discharges. On the contrary, the depolarization process of CSD is a loss of excitability state lasting minutes and very similar to the ischemic depolarization resulting in loss of function [2, 3, 13]. Whole-cell recordings in vivo confirm the prolonged depolarization up to 345 s with neuronal silencing during CSD, which was followed by reduced spontaneous synaptic activity and AP firing. Significant reduction in the spontaneous excitatory and inhibitory postsynaptic potentials, yielding an imbalance towards inhibitory tone, lasts longer than 90 min [13].

Consequently, the initial phase of CSD where the membrane resistance is lost and membrane potential is seized in depolarized state is a complete electrical silence that is unlikely to be associated with excitation of any cortical neuron leading to positive aura symptoms. On the other hand, a shower of population spikes lasting 2–4 s within the late gamma band that precedes DC drop [3, 14] could be attributed to positive symptomatology. However, the brief duration of antecedent spike shower may not be perceived consciously and may not be manifested as a clinical symptom. Even if it is perceived as a positive symptom, it has to be overwhelmed by following long-lasting negative symptomatology. Thereby, even the expectation of increased activity of excitatory cortical neurons with positive symptomatology alone is not compatible with basics of canonical CSD.

CSD was employed to model the depressed neuronal activity in the cerebral cortex and never been used to produce a brain hyperexcitability model. Transient loss of function of the cerebral cortex as a result of CSD was also noticed by other researchers and used as a method to generate functional decortication [15, 16]. The fact that CSD was used to produce functional decortication/ ablation [17, 18] indicates its massive negative functional consequences in mammalian brain. Recent studies unveiled the occurrence of SD waves in human cortex immediately before brain death, malignant ischemic stroke, aneurysmal subarachnoid hemorrhage or intracerebral hemorrhage and traumatic brain injury and their association with increased mortality and morbidity [4‐8]. The latter conditions differ from migraine not only due to a severe disability but also the fact that majority of migraine auras consist of positive symptoms in visual, and/or somatosensory modalities.

The trigeminal nerve mediates the pain sensation from cephalic structures in the anterior portion of the head and plays a crucial role during headache phase of migraine. Peripheral trigeminal nerve endings arise from dural vasculature and therefore are termed as the trigeminovascular system. Upon activation, trigeminal nociceptors release their contents such as calcitonin gene-related peptide (CGRP) to perivascular space [11, 12]. Stimulation of trigeminal neurons at the ganglia, in experimental animals, results in neurogenic inflammation characterized by the release of CGRP and substance P, vasodilation, increase in blood flow and protein extravasation in the dura mater. Therefore, any activation of meningeal perivascular trigeminal nociceptors by potassium, arachidonic acid or other molecules released during CSD, could initiate signal transmission orthodromically to the pain processing structures, while inducing antidromic neurogenic inflammation in the dura mater in experimental animals [12].

In rat brain‚ CSD in pain-insensitive brain parenchyma was related to unilateral activation of trigeminal afferents and subsequent activation of 2nd order trigeminal pain neurons. The possible link between CSD and head pain was supported by the following findings in anesthetized rodent experiments; 1) Two dimensional blood flow imaging of both the cerebral cortex and the dura mater revealed that following CSD wave, blood flow in the middle meningeal artery (MMA) notably increased for an hour in a trigeminal nerve dependent manner [12]. MMA was a perfect dural structure to observe the trigeminovascular system due to its dense innervation by trigeminal nerve. The increase in the ipsilateral MMA blood flow peaked 20 min after CSD and coincided with oligemic phase of the brain [12]. 2) CSD triggered trigeminal nerve dependent plasma protein leakage, prominent around MMA within overlying dura mater in the ipsilateral hemisphere [12, 19]. 3) CSD induced expression of the immediate early gene product c-FOS in trigeminal nucleus caudalis (TNC) in the ipsilateral brainstem [12, 19, 20]. Activation of the trigeminal nucleus also initiated trigemino-parasympathetic reflex which led to dural vasodilatation and hyperemia via release of vasoactive intestinal peptide, nitric oxide (NO) and acetylcholine, 4) CSD could activate perivascular trigeminal nerve endings [21]. Those studies were conducted in rodents under anesthesia and suggested a link between an intrinsic brain event and lateralized head pain through peripheral bipolar neurons of trigeminal ganglia. Though the latter was disputed by Lambert et al. [22], where authors showed the persistence of CSD induced TNC neuronal activation following trigeminal nerve transection and thereby suggested a direct central pathway. Possible involvement of central route to activate TNC was also proposed to be operational particularly in awake subjects through thalamus and other brainstem structures [23]. CSD has also been shown to result in bilateral TNC activation [24, 25], decrease in mechanical pain thresholds in the head and pain-related behavior [25, 26].

CSD experiments in rodents showed the opening of pannexin 1 megachannels, MMP activation and BBB breakdown providing the possible mechanisms that explain how the intracellular nociceptive and vasoactive molecules induce neuroinflammation and find a way to pass through the blood-brain barrier and reach to the perivascular trigeminal nociceptors in the dura mater [19, 27, 28].

Another supporting evidence for the linkage of CSD to head pain came from the drug studies in rodents. Trigeminal activation at the second order neurons in TNC by CSD was reversed by sumatriptan [12, 20], valproic acid [23], topiramate [29], CGRP receptor antagonists [30] or metoclopramide [31]. Latter two agents also significantly reversed the CSD induced pain behavior such as grooming and freezing in addition to reduced von Frey thresholds and cold allodynia in freely moving rodents. It is important to note that acute use of aforementioned pharmaceutical agents drugs do not inhibit or influence CSD waves.

CSD is capable of inducing thalamic reticular nuclei (TRN) which is a key structure as a gate keeper in thalamocortical information flow and plays a role in selective attention, sensory discrimination, lateral inhibition, and sleep. TRN activation was in the visual sector and SD waves could propagate to TRN if the rats are devoid of any effect of anesthesia. The propagation of SD waves to TRN was blocked by valproic acid administration [23]. Therefore, its involvement in CSD is a very important step to look from a broader perspective into the cortical function since the cortical processing of any somatosensory, motor or cognitive function includes the thalamus, an inseparable counterpart of the cortex.

Transgenic mice expressing mutations associated with familial hemiplegic migraine (FHM) showed increased CSD susceptibility, where the glutamatergic synapse was the common target [32]. Reduced threshold for CSD induction and an increased CSD propagation speed were detected in both Cacna1aR192Q knock-in mice that carry the FHM1 mutation and FHM2 knock-in mice carrying the human W887R mutation in the ATP1a2 orthologous gene. Behavioral consequences of CSD in those mice exhibited several features attributed to migraine head pain and photophobia. However, their relation to trigeminovascular system or pain nucleus was not studied in mutant mice. Cacna1aR192Q knock-in mice displayed signs of photophobia and chose to remain in open space to avoid light in the bright closed arms condition and they spent significantly more time in the dark open arms than wild-type (WT) mice in modified elevated plus maze. Head grooming was increased in R192Q mutant mice and number of long strokes initiating in the oculotemporal region were significantly higher in Cacna1a transgenic mutants than WT animals showing a possible site of pain. Mutant mice also showed increased eye blinks associated with a whole-body shuddering behavior. Abnormal head grooming and eye behaviors were reversed by rizatriptan, a selective 5-HT1B/1D receptor agonist, an acute antimigraine medication in a dose-dependent manner [33]. In another study, Cacna1a knock-in mutant mice showed a baseline pain face with higher baseline mouse grimace scores compared to WT mice, which was prevented by intraperitoneal administration of 50 mg/kg rizatriptan [34].

CSD could explain some features of migraine but following unresolved points needs to be addressed.

Hence the concept of canonical CSD waves that propagate strictly within the human cerebral cortex and cause aura symptoms and headache, has several shortcomings to justify manifestations of the migraine disorder, additional mechanisms and views are needed.

CSD could still occur in a restricted area in the migraineur’s cortex where the propagation of depolarizating waves are easily stopped and terminated, as shown in human cortex. However, the consequence of an unresponsive cortical focus should be taken into account in a context with thalamocortical oscillations and widespread brain network systems. Understanding of cerebral cortical function requires a knowledge of complex interaction of the thalamocortical network. The thalamus sends projections to almost all cerebral cortex, and the cortical region receiving projections from the thalamus, sends outputs to one or more thalamic nuclei, in return [119‐121]. In that sense the thalamic nuclei can be considered as a first-order and higher-order relay nuclei. First-order thalamic nuclei, such as the lateral geniculate nucleus (LGN) and the ventral posterior (VP) nucleus, receive and relay inputs from ascending visual and somatosensory pathways to first order cortical areas [119, 120]. On the other hand, higher-order thalamic nucleus, receive inputs principally from the 5th layer of the cerebral cortex and relay information between different cortical areas [120]. Thereby, transthalamic cortico–cortical circuits can synchronize oscillations over cortical distances [18, 121].

Thalamus and cortex are in a continuous interaction in processing any information particularly related to sensory, motor, language, cognition and consciousness. Abrupt silence of cortical counterpart as a result of SD would result in an imbalance that needs to be filled/adjusted with the change of usual firing pattern of thalamic nuclei. CSD is estimated to be induced if a stimulus simultaneously depolarizes a minimum critical volume of 1 mm³ in rodent cortical brain tissue [122]. Even if its propagation is blocked, such a cortical focus is sufficient to create a change in ongoing cortico-thalamic drive on thalamic relay nuclei and thalamic reticular nucleus. For instance, sudden loss of excitatory drive on GABAergic TRN that inhibits other thalamic nuclei, would result in increased sensory excitation, reduced discrimination and reduced lateral inhibition. The latter findings were shown in migraine patients [123, 124]. In episodic migraine patients, a transient but robust prolongation in somatosensory temporal discrimination thresholds (STDTs) was shown during the migraine attacks whereas STDTs were normal in tension-type headache patients [123, 125]. The discrimination of the exact entry of sensory stimuli was disrupted in migraine attack, indicating an impairment in central somatosensory and cognitive processing [126].

The pulvinar in the dorsal thalamus contribute to the visual information processing and multimodal sensory integration [102]. A dysfunction in higher order thalamic relay nucleus such as pulvinar dysfunction may be related to positive visual hallucinations superimposed to a primary incoming information and could be more compatible with the patients’ visual auras. High levels of connectivity of pulvinar with the primary motor-sensory cortices, superior parietal gyrus, precuneus and visual cortex [127] are in favor of the latter view. Recently impaired sensorimotor integrity and reduced cortico-cortical inhibition between somatosensory and motor cortices were shown to be associated with migraine attacks [128]. Notably, facilitation of motor responses to a conditioned sensory stimulus was detected even several hours prior to headache onset. Therefore, the contribution of higher order thalamic nuclei to these changes seems irrefutable.

It is also probable that SD ignition in the cortex could induce SD in TRN [23], as subcortical structures are still relatively more vulnerable to SD compared to highly convoluted cerebral cortex in humans [129]. Majority of somatosensory aura symptoms are cheiro-oral type (paresthesia in the fingertips and perioral area, tongue and palate). Majority of cheiro-oral syndromes has been associated with thalamic or brain stem lesions [130]. Likewise, in the thalamic ventral posterior (VP) nucleus, the representation of the acral parts of all digits, the tongue and the lips in close proximity, is more compatible with paresthesia starting from distal parts of all digits moving proximally skipping upper arm, shoulder, head, forehead, eyes and nose [131]. Additionally, aphasia can also occur as a result of thalamic dysfunction [132] and thalamus was implicated in migraine [23, 133, 134]. Pontine and other brain stem projections from cholinergic and noradrenergic nuclei, LC to the thalamus provide multiple modulatory drive on thalamo-cortical interactions. Thalamic player in accordance with the cortex would enlighten positive aura symptoms and concurrent manifestation of different symptoms related to certain cortical areas such as visual, somatosensory, language and motor cortices connected through thalamic hub.

Migraine headache is a heterogeneous complex disorder and it seems implausible to propose single mechanism to elucidate sufficiently all clinical manifestations. Therefore, more than one mechanism and multiple locations must be considered in the development of migraine headache. While the canonical CSD could explain visual aura and following headache in certain population of patients, it has flaws in reasoning multiple sensory auras. It is also important to account thalamus as a hub that provides extensive connections among the visual, somatosensory, language and motor cortical areas. Emergence of SD in reticular nucleus or other locations in the thalamus could thereby rationalize some features of migraine. Additionally, including central pathways activated by thalamocortical dysfunction in the process of headache development would improve our understanding of migraine.