Suppressive effect of SD on spontaneous cortical activity of awake and anesthetized animals

The present study shows that vulnerability of slow and fast cortical oscillations to suppressive effects of SD profoundly differs, especially in awake state. In freely behaving rats, SD is accompanied by strong cessation of fast cortical oscillations, particularly pronounced in the gamma (25–50 Hz) range, while the slowest (delta, 1–4 Hz) activity was not depressed during SD but even increases afterwards. The pattern of partial ECoG depression may explain incomplete cessation of cortical activity during SD previously reported in awake rats and rabbits [17, 18] and the failure of most clinical studies to detect clear EEG depression during migraine aura in conscious patients [7, 8]. Similar to our experimental findings, MEG/EEG data from migraine patients reported (1) ipsilateral suppression of high-frequency (alpha and gamma) cortical activity during visual aura that was suggested to contribute to inhibition of visual function and phosphene generation [31] and (2) increased delta power in the occipital cortex (posterior slow waves) during typical migraine aura [32] and FHM aura [33].

Introduction of anesthesia, despite its rather mild effect on parameters of dc shifts associated with SD, significantly modifies the pattern of SD-induced ECoG depression, weakening suppression of gamma oscillations and intensifying depression of cortical activity in other frequency bands. Under anesthesia, SD is accompanied by wideband suppression of cortical activity with the strongest power drop of slow delta activity and milder reduction of faster oscillations. The result is in line with clinical data obtained in sedated patients with traumatic brain injury in which EEG/ECoG suppression during SD was mainly determined by suppression of slow cortical activity in the delta frequency band (reduction to 47%) while high-frequency oscillations were less depressed [9]. Intense wideband depression found during SD under anesthesia in our study corresponds well to SD-induced complete ECoG depression previously described in anesthetized rabbits, rats, mice and pigs [4, 5, 10].

In both awake and anesthetized states, drop of ipsilateral cortical oscillations power always peaks during DC-shifts confirming the well-known idea that the most prominent deactivation of the cortex occurs during depolarization phase of SD as a result of depolarization block of neuronal activity [4]. However, multiple preclinical evidence show that SD-induced suppression of spontaneous cortical activity lasts significantly longer than the depolarization phase of SD ([e.g., [5]). Our results are in line with the well-known data and show that duration of the ECoG depression may depend on vigilance state, cortical region and type of cortical oscillations.

Remote effects of SD on high-frequency cortical activity of awake brain

The striking feature of SD effects on cortical activity of awake animals was bilateral depression of high-frequency gamma oscillations induced by unilateral SD. A decrease in gamma power was observed both in the cortex invaded by SD and in the unaffected contralateral cortex with region-specific time courses – short-lasting depression in the occipital cortex and long-lasting early-onset decline in the frontal cortex. Bilateral suppression of alpha band (8–11 Hz) cortical oscillations was described following KCl-induced unilateral cortical SD that was referred as a diaschisis manifestation [34]. Our study shows that the SD diaschisis selectively involves fast cortical oscillations and exhibits state- and region-dependent features. Recently, we reported that in awake rats a single unilateral cortical SD elicited a transient loss of interhemispheric functional interactions, especially pronounced in the beta-gamma frequency bands [35]. The functional decoupling may underlie the ECoG depression produced by contralateral SD.

In the frontal cortex of awake rats, beta and gamma band power began to reduce soon after SD initiation in the BLA and progressively diminished with SD approach to the cortex (during the first two minutes post-injury). Similar depression of beta cortical activity starting long before SD appearance at the recording sites was reported in patients with traumatic brain injury, in which SD occurrence was closely associated with reduced beta band power [36]. Given that the gamma depression preceding cortical SD was absent after sham stimulation not triggering SD and that during the early period SD traveled over the remote subcortical and cortical regions [18, 20,21,22] (see Fig. 1), we conclude that the early-onset cessation of fast activity was produced by network effects of SD invading distant brain regions. Previously, it has been shown that neuronal (unit) activity and sensory evoked responses of the cerebral cortex were reduced during subcortical (striatal and thalamic) SDs [37, 38]. That is, subcortical SD can alter cortical function by transient elimination of afferent inputs to the cortex and functional disconnection of the cortex from deep brain regions during the depolarization phase of subcortical SD [6, 16]. However, recent experimental evidence indicate that distant effects of SD may be more complex. In awake mice, cortical SD has been reported to elicit transient neuronal activation of the ipsilateral thalamus [39].

The frequency-, state- and region-specific character of the remote effects of SD may explain why it usually remained unnoticed in experimental studies. Also, in most studies SD was initiated within the rodent cortex, the small size of which hampered detecting the distant effects of SD. A role of initiation site localization (the parieto-occipital cortex in most studies and the amygdala in the present work) cannot be excluded. Our experimental design with initiation of SD in remote extracortical region and a significant time lag between SD induction and its arrival to recording points mimics better SD traveling over long distances such as those observed in the human cortex.

It remains unclear why the remote effects of SD are strongly expressed by the frontal cortex. In the cortical region, gamma-band depression preceded SD arrival and involved both affected and unaffected hemispheres (Fig. 4). Urethan anesthesia abolished the early pre-SD and contralateral gamma depression (Fig. 5). Similarly, thalamic activation during cortical SD was eliminated by anesthesia [39]. Anesthetics are known to diminish activity of brainstem arousal nuclei and affect bidirectional communication across the brainstem, thalamus, and cortex. The frontal cortex receiving robust ascending projections from arousal- and pain-modulation brainstem nuclei [28, 29] may be particularly sensitive to the changes in cortico-subcortical interactions. Also, the vulnerability of the frontal cortex may result from its contiguity to subcortical pathways of SD traveling from the amygdala (Fig. 1) that implies the existence of spatial limits for the remote effect expression. At last, anatomical/functional connections between sites driving the remote SD effects and the two cortical regions may differ. The frontal cortex is the most important recipient of a direct input from the periaqueductal gray matter (PAG) while occipital cortex receives only a minor PAG projection [28].

Migraine is a disorder of cortico-subcortical interactions. It is thought that activation of subcortical structures drives symptomatology of premonitory and headache phases of the migraine attack while cortico-thalamic events are accepted to determine sensory manifestations of the aura phase [1, 2, 14, 26]. Cortical SD has shown to invade the visual domain of thalamic reticular nucleus [14] and to activate thalamic ventral posteromedial nucleus [37], which are both relevant to sensory information processing. Aberrant activity of brainstem arousal and nociceptive networks during premonitory period is suggested to initiate migraine attacks [1, 2, 26]. Hyperexcitation of ascending subcortico-cortical pathways can trigger cortical SD in awake animals [40, 41]. On the other hand, SD involving subcortical structures is also referred as a plausible mechanism for some aura symptoms in patients [3, 6, 14].

The present study shows that in awake conditions SD exerts remote effects on fast activity of the cortex and this effect is abolished by urethan anesthesia. This suggests that SD occurring in the conscious brain of migraine patients exerts not only direct local ECoG depression in the affected tissue but may also produce indirect suppression of high-frequency gamma oscillations in distant brain regions.

Effect of a single unilateral SD induced in the amygdala on spontaneous behavior

The present study shows that traveling SD from BLA to the cortex is reliably accompanies by episodes of forced circling, freezing behavior and “chewing” movements. As shown previously, circling behavior time-locks SD invasion of the striatum [18]. Its reproducible occurrence soon after the BLA pinprick indicates regular propagation of SD initiated in the amygdala to the striatum. Association of cortical SD with freezing behavior has been reported previously [14, 16, 18]. It has been suggested that mechanisms of the SD-related freezing involve the amygdala playing the critical role in expression of the fear and anxiety behavior [16]. Our findings support the idea and show that recurrent episodes of freezing appear immediately after SD initiation in the amygdala. In the present study, new behavioral pattern associated with SD – recurrent masticatory jaw movements – was identified. Given that the behavior is generated by trigeminal circuits controlling orofacial motor function [42], the SD-associated “chewing” movements may indicate activation of downstream nociceptive pathways during SD propagation in the brain.

Relevance to pathogenesis of migraine aura

Migraine aura is a neurologic condition characterized by transient visual, somatosensory and language symptoms that develop before headache phase of migraine attack. Cortical SD induced in experimental animals represents a highly translational model of the acute neurological deficit. Though experimental SD recapitulates many characteristics of migraine aura in human subjects, some features of SD do not match well clinical pattern of aura [3]. Our study shows that the mismatch may be related, at least partially, to the fact that the main body of our knowledge about electrographic characteristics of SD has been obtained in anesthetized animals. Here, we found that SD elicits more complex changes in cortical activity in the awake state compared to those observed under anesthesia. Some of the changes detected only under awake condition may underlie several unexplained features of migraine aura.

First, bilateral aura symptoms are frequently observed in migraine patients but mechanisms of the aura pattern remain unclear based on properties of unilateral SD described in anesthetized animals (depression is confined to the cortex affected by SD). Multiple experimental studies, including the present one, showed that under anesthesia suppressive effect of unilateral SD is confined to the ipsilateral cortex. Here, we found for the first time that in awake conditions the contralateral cortex unaffected by SD also shows transient depression of cortical gamma oscillations. It is known that high-frequency cortical activity plays the critical role in processing of sensory information and impaired regulation of the activity is referred as a hallmark of neurological dysfunction. The ability of unilateral SD to produce in awake brain reversible bilateral depression of gamma oscillations may potentially underlie bilateral sensory disturbances during migraine aura.

Second, visual and somatosensory aura symptoms can appear in rapid succession or simultaneously. Such symptomatology cannot be explained by direct traveling SD over the human cortex due to a long distance between the visual and somatosensory cortical regions. Moreover, functional imaging studies did not find such propagation patterns in patients and showed that the event underlying visual aura propagates along a single gyrus or sulcus [43]. Based on the clinical data, multifocal triggering cortical SD during aura has been suggested [2]. Our study revealed that in wakefulness beta-gamma depression spreads beyond a spatially limited SD event and produces ECoG depression in broader cortical areas not invading them. Restricted traveling SD along the gyrus/sulcus thus can drive visual aura and exert distant effect on activity of the somatosensory cortex, yielding several sensory symptoms simultaneously. Given an important role of high-frequency gamma oscillations in the frontal cortex in network-level computations, their prolonged depression produced by SD in awake brain may underlie cognitive impairments during migraine attacks.

Finally, the majority of migraine patients exhibit positive sensory symptoms which remain unexplained based on mainly suppressive effect of SD on cortical activity. Previously, we have shown that in awake rats cortical SD is followed by transient hyperexcitation of the ipsilateral cortex [35]. The present study confirmed the finding and showed that in awake state SD is followed by increased delta power in the occipital cortex. It can be speculated that the post-SD activation of the visual cortex may be perceived as positive aura symptoms.

The strength of the present study was reliable induction and recording of SD in freely behaving animals that mimics better conditions of migraine aura in patients. Further, detailed investigation of temporal evolution of cortical activity following SD is important advantage of the study. Limitations include small groups of animals that resulted from difficulties of obtaining long artifact-free ECoG recordings in freely behaving rats, and low spatial covering of SD propagation. The lack of direct electrographic evidence of SD occurrence during migraine attack in patients complicates translation of the experimental results to humans. Pathways of the non-synaptic propagation of SD over the lisencephalic cortex of rodents may differ from those in the gyrencephalic cortex of humans. Non-uniform velocity of SD propagation in gyri and sulci [44] is distinct from the constant rate of SD expansion across the lisencephalic cortex of rats. Complex spatiotemporal patterns of SD spread, including spiral and reverberating waves, seem to be more common in the gyrencephalic cortex [45].

To sum up, our study shows that slow and fast cortical oscillations exhibit pronounced difference in their vulnerability to suppressive effect of SD. In conscious drug-free brain, high-frequency gamma oscillations involved in sensory and pain processing are particularly sensitive to SD influence and show spatially broad long-lasting cessation. Why gamma activity playing the critical role in the function of the conscious brain and pain perception is more vulnerable to suppressive effects of SD in awake conditions remains unclear and needs further investigation. The state-dependent features of transient cortical dysrhythmia induced by SD should be considered in translation of experimental data to clinic of migraine and understanding pathophysiological mechanisms of migraine aura.