An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Noebels JL, Avoli M, Rogawski MA, et al., editors. Jasper's Basic Mechanisms of the Epilepsies. 5th edition. New York: Oxford University Press; 2024. doi: 10.1093/med/9780197549469.003.0062
Spreading depolarization (SD) is a massive, slowly migrating wave of near-complete cellular depolarization of neurons and glia associated with profound cytotoxic edema. SD is frequently generated following acute brain injury and contributes to the transient and chronic deterioration of neurological conditions. Clinical and basic studies also suggest that SD is a significant contributing mechanism of headache and neurological deficits in some migraine with aura. Because of the increasing evidence supporting the role of SD in these clinical conditions, there is a growing interest in exploring other neurological disorders that involve SD, including epilepsy. Both basic and clinical studies repeatedly demonstrate that seizure and SD can be co-generated in a temporally and spatially associated manner, suggesting the possibility that SD may be a comorbid feature of epilepsy. On the other hand, some studies suggest that the chronically epileptic brain is resistant to SD, and thus the relationship is still unclear. For technical reasons, clinical evidence of spontaneous SD generation in epilepsy and its interaction with seizures is very limited. Because mechanisms of SD and seizures can vary significantly in different contexts, their potential interaction likely requires individual evaluation depending on the type of epilepsy. This chapter will review the similarity and distinctions of molecular and physiological mechanisms of seizures and SD, and then discuss their clinical interaction in different pathological settings.
Spreading depolarization (SD) is a neurophysiological phenomenon characterized as a massive cellular depolarization associated with profound cellular volume changes which slowly spreads through brain grey matter, typically at a rate of 2–9 mm/min (Dreier, 2011). The term “spreading depolarization” began to be used relatively recently, and until then, these events were named mechanistically, for example, spreading depression, hypoxic spreading depression, anoxic depolarization, and ischemic depolarization, depending on the context (Somjen, 2001). The integration of these phenomena helps discussion of their shared mechanisms as they often appear as intermediate and/or partially overlapping, rather than fully segregated, phenomena in many clinical and experimental settings. However, it also created ambiguity and confusion because the underlying mechanisms of SD can significantly vary depending on the experimental model and pathological context. SD can be better understood as a continuum in that propagation mechanisms and consequences of a single SD wave can continuously change depending on the pathophysiological condition of the affected tissue conditions (Dreier & Reiffurth, 2015). Similar confusion is seen in the umbrella term “seizure” or “epileptic activity,” that are used for almost any pathological discharge regardless of underlying mechanisms. For better understanding of mechanistic relationship between seizure and SD, their interaction should be carefully analyzed individually, rather than overly generalize the mechanisms.
SD can be generated in diverse context, however, there are features that characterize SD as a unique neurophysiological phenomenon. This section will introduce the basic characteristics of SD initiation and propagation (see Fig. 62–1).
KCl-evoked SD generated in hippocampus CA1 in vitro. A. Whole-cell recording in CA1 pyramidal neurons. SD is detected as a slow near-complete membrane depolarization associated with a profound decrease of membrane resistance. B. Extracellular recording. (more...)
SD is characterized as a self-regenerative feedforward event originating from a priming local tissue depolarization. the initiating depolarization event must be severe enough to elevate and sustain extracellular excitatory solutes (such as K+ and glutamate) to the supra-physiological level in a sufficiently large volume of tissue. This severe membrane depolarization is associated with the intracellular accumulation of Na+ and compensatory transmembrane Cl− influx and K+ efflux (Dreier & Reiffurth, 2015; Somjen, 2001). The ion redistribution due to the near complete loss of ion gradient result in the imbalance of extra/intracellular osmolality which drives imblance, resulting in cytoplasmic swelling and extracellular space constriction (i.e., cytotoxic edema; Andrew et al., 2007). During SD, the fraction of tissue extracellular space decreases from 10%–15% to ~5% (Jing et al., 1994; Vargová et al., 2001), and this volume change is an important contributing mechanism of supra-physiological elevation of the extracellular solutes.
The excitatory solutes released during the initial depolarization slowly diffuse and depolarize the surrounding tissue. When the depolarization reaches a threshold level and the excitatory solute concentrations within the surrounding tissue also elevate to the supra-physiological level, resulting in regenerative feedforward cellular depolarization that slowly spreads through brain tissue at a rate of 2–9 mm/minute. In this regard, SD is a “regenerative volume transmission” wave of excitatory solutes. Because of the propagation mechanism, SD is usually confined within a physically continuous gray matter, and thus SD normally does not spread to anatomically separated structures (Fig. 62–2). This is in contrast to the propagation of seizure which spreads via neural synaptic circuitry, allowing extremely fast propagation and spread into distant brain regions. The anatomical difference in seizure and SD susceptibility will be further discussed later.
Neurophysiological changes associated with seizure and SD propagation. A. A seizure initiates as synchronous recurrent discharges in recurrently connected excitatory networks which self-sustain and spread to synaptically connected neurons. Seizure activity (more...)
During SD, neurons near-completely depolarize for a minute or longer. This slow massive tissue depolarization generates the iconic extracellular negative DC potential shift of SD. The sustained loss of transmembrane ionic gradient during membrane depolarization, especially intracellular Na+ loading, rapidly activates the Na+-K+-ATPase and accelerates ATP consumption. The concomitant Ca2+ overload impairs mitochondrial function resulting in additional energy burden. Such intracellular ATP depletion is important mechanism for the unusually prolonged and severe disruption of intra- and extracellular ion gradients. The repolarization phase is a key determinant of the consequence of SD, and prolongation or failure of repolarization results in deleterious consequences. Similar intracellular ATP depletion, slow membrane depolarization, and channel inactivation can also occur in seizures; however, these neuronal exhaustions are milder and linked to seizure termination rather than tissue propagation (Lado & Moshé, 2008).
The sustained membrane depolarization inactivates most voltage-gated ion channels and creates a transient depression of spontaneous electroencephalographic (EEG) activity in the affected tissue (Ichijo & Ochs, 1970; Phillis & Ochs, 1971). The prolonged EEG suppression may also involve other neuromodulators such as adenosine receptor-1 (Lindquist & Shuttleworth, 2017). The SD-induced EEG depression is not a complete silencing, but often seen as a frequency-dependent amplitude attenuation. The severity and frequency dependency of EEG depression may be affected by the recording region, recording methods, and brain status (e.g., awake vs. anesthetized, cerebral perfusion level, tissue injury, etc.) and, in some cases, a part of EEG spectral band activity can even be increased. The complex EEG activity changes during SD may reflect in part heterogeneous cellular depolarization. At cellular level, single-cell electrophysiological studies suggested that the neuronal response to SD can vary significantly from cell to cell, and some neurons may even not participate (Czéh et al., 1993; Sugaya et al., 1975). In contrast cellular discharge pattern during seizure is highly heterogeneous depending on the seizure type and phase.
Among the excitatory solutes, extracellular K+ is considered as the most important mediator of SD generation/propagation, and theoretically, sufficiently high K+ alone can create and maintain the feedforward depolarization. Despite its critical role, the underlying molecular mechanisms serving as the source of extracellular K+ are not fully understood. Some K+ channels directly contribute (Aitken et al., 1991; Y. Xie et al., 1995), while nonselective cation channels such as NMDA-type glutamate receptors (NMDARs) and ion exchangers may also contribute depending on the context. In addition, extracellular K+ concentration rises as a result of necrotic cell death, and an estimate suggests that K+ concentration can reach 40 mM in tumor interstitial fluid (Eil et al., 2016). Similar extracellular K+ concentrations may be achieved in severe brain injury, where massive necrotic cell death and hemolysis are present.
Seizure activity also elevates extracellular K+ concentration, but at a lower level than in SD. During the seizure, K+ concentration quickly reaches ceiling levels of around 10–14 mM and stays nearly constant until the seizure terminates (Dreier & Heinemann, 1991; Fisher et al., 1976; Hablitz & Heinemann, 1989; Raimondo et al., 2015). In some conditions, exogenously elevating the extracellular K+ above 8 mM can provoke epileptic discharges in vivo and in vitro (M. S. Jensen & Yaari, 1997; Korn et al., 1987; Traynelis & Dingledine, 1988; Zuckermann & Glaser, 1968). However, extracellular K+ elevation alone is unlikely the general trigger of seizures; instead, it is considered as a consequence of discharge that helps sustain excitability.
In addition to the peak concentration, there may be a distinction in the subcellular compartment responsible for K+ release during seizure and SD. K+ elevation during a seizure is more pronounced in the somatic layer than the dendritic field (Fisher et al., 1976), while the extracellular K+ surge during SD is similar in both structures (Somjen & Giacchino, 1985). The higher peri-somatic K+ concentration in seizures suggests that the majority of K+ is released during recurrent action potential discharges, while in SD additional sources contribute to the K+ elevation.
Extracellular K+ elevation is one of the common contributing mechanisms of SD initiation and propagation. However, in healthy brain tissue, extracellular K+ surge is often small or even absent before SD arrival, and it alone is likely insufficient to drive the depolarizing wave. In many instances, additional contributory mechanisms independently or synergistically create and maintain the feedforward depolarization, and the difference in the contributing mechanism result in distinct pharmacosensitivities. In general, the depolarizing conductance during SD is the persistent inward sodium current, and any channels capable of generating such a conductance may be able to contribute to SD. This section will discuss contributing mechanisms with a focus on those relevant to both SD and seizures. There are contributing mechanisms not covered here, and a more systematic review of pharmacosensitivity of SD is available elsewhere (Klass et al., 2018).
NMDA-type glutamate receptor (NMDAR) activity is an important contributing mechanism for many forms of SD (Klass et al., 2018). NMDARs have extremely slow inactivation kinetics (seconds) compared to the AMPA and kainate-type glutamate receptor channels that inactivate in tens of milliseconds, and this exceptionally slow channel inactivation kinetic is suitable for generating the initial sustained depolarization phase (Lauritzen & Hansen, 1992; Nellgård & Wieloch, 1992). NMDAR activation may also contribute to part of the extracellular K+ elevation due to its intrinsic K+ conductance that can be revealed under reduced extracellular Na+/Ca2+ (Ichinose et al., 2003) conditions, or in concert with the coupled BK-type K+ channels (Isaacson & Murphy, 2001; J. Zhang et al., 2018). In some experimental settings, NMDARs expressed at presynaptic terminals contribute to regenerative glutamate release (Zhou et al., 2013). In addition to SD generation/propagation, NMDAR activation during SD can also influence the injurious consequences of SD, since sustained activation counteracts membrane repolarization, and prolong SD duration (Aiba & Shuttleworth, 2012; Mei et al., 2020).
In contrast to their efficacy on SD, NMDAR antagonists are usually not the best candidate for antiseizure prophylaxis and are reserved for emergent use in some cases of refractory status epilepticus (Kramer, 2012). NMDAR inhibition can occasionally increase neuronal excitability (Hou & Zhang, 2017; Tatard-Leitman et al., 2015), and exacerbate psychiatric side effects (Löscher & Hönack, 1993; T. Su et al., 2018). On the other hand, AMPA receptor antagonists have an antiepileptic effect, and a noncompetitive AMPA antagonist, Perampanel, is approved to use for some epilepsy (Hanada et al., 2011).
The level of available NMDAR agonist, glutamate, influences SD propagation. extracellular glutamate imaging study reported a surge of extracellular glutamate starting a few seconds after the DC shift in vivo (Enger et al., 2015). On the other hand, recording excitatory postsynaptic currents suggests the synaptic glutamate release is increased a few seconds before the arrival of SD in vitro (Aiba & Shuttleworth, 2012) and glutamate concentration within the synaptic cleft may be elevated and contribute to initial depolarization. In many instances, glutamate is primarily released via synaptic Ca2+-dependent vesicular exocytosis. The PQ type-Ca2+ channel (Cav2.1/CACNA1A) is especially important in mammalian brain, and modulations of channel function bidirectionally modulate the SD susceptibility; Gain-of-function mutations in the CACNA1A are linked to familial hemiplegic migraine type-1 (FHM-1), and some FHM-1 mutations increase SD susceptibility in mice (van den Maagdenberg et al., 2004, 2010). In contrast, a loss-of-function PQ-channel mutation decreases glutamate release and reduces SD susceptibility (Ayata et al., 2000). Acute pharmacological inhibition of P/Q type channel by ω-agatoxin also inhibits some forms of SD (Kunkler & Kraig, 2004; Richter et al., 2002), and the nonspecific sodium channel blocker lamotrigine, that partially inhibits this Ca2+ current, can also inhibit some forms of SD (Bogdanov et al., 2011; Eikermann-Haerter et al., 2015; Stefani et al., 1996). In addition to the PQ-type Ca2+ channel, other presynaptic Ca2+ regulators and vesicle fusion machinery may also modulate SD susceptibility. Inhibition of N-type Ca2+ channel (Cav2.2/CACNA1B) (Richter et al., 2002) or depletion of vesicular glutamate by bafilomycin (Zhou et al., 2013) inhibits SD, while a leaky RyR2 mutation that causes aberrant cytoplasmic Ca2+ elevation lowers SD threshold (Aiba et al., 2016). Reverse glutamate transporter was suggested as a mechanism of extracellular glutamate release in an in vitro anoxic depolarization model (Rossi et al., 2000); however, its contribution to SD initiation/propagation seems to be limited.
Some voltage-gated sodium channels generate a persistent inward current and can contribute to the initial depolarization of some forms of SD. Tetrodotoxin (TTX), a potent selective voltage-gated Na+ channel blocker, attenuates cellular depolarization of layer 2/3 cortical neurons during SD in unanesthetized mice (P. M. Sawant-Pokam et al., 2017) and partially inhibits hypoxia-induced SD in hippocampal slices in vitro (Aitken et al., 1991; Müller & Somjen, 2000). Lamotrigine, a persistent sodium channel blocker (also Ca2+ current inhibitor as discussed above), and dibucaine, a use-dependent channel blocker, also inhibit SD triggered by KCl or SD generated by simulated ischemia in vitro (Bogdanov et al., 2011; Douglas et al., 2011; Eikermann-Haerter et al., 2015). On the other hand, some studies did not detect inhibitory effects of TTX on SD tested in hippocampus, retina, or cerebellum preparations (Aiba & Shuttleworth, 2012, 2014; Kow & van Harreveld, 1972; Tamim et al., 2021; Tobiasz & Nicholson, 1982). The variable sensitivities suggest that the contribution of voltage-gated Na+ current is not generally required for SD, and its contribution varies depending on the SD model and/or brain structures tested.
Various Na+ channel blockers are used as antiepileptic drugs, while some local anesthetics, such as lidocaine and bupivacaine, paradoxically induce seizures when rapidly given at high doses via intra-arterial routes (D. L. Brown et al., 1995; DeToledo et al., 2002). It has been suggested that these use-dependent blockers preferentially inhibit fast-spiking inhibitory neurons, thereby increasing the network excitability.
Voltage-gated Na+ channels are encoded by the SCN gene family and mutations in the sodium channel α-subunit, SCN1A/Nav1.1, are associated with genetic migraine and/or epilepsy depending on the mutation type. Nav1.1 is expressed predominantly in inhibitory neurons in the hippocampus, but, in the neocortex, these Na+ channels are also present in subgroups of excitatory neurons (Yamagata et al., 2021). Pathogenic gene mutations identified in Nav1.1 in the familial hemiplegic migraine type-3 (FHM-3) patients are proposed to facilitate a K+ surge by preferentially increasing the excitability of inhibitory neurons (Chever et al., 2020; Desroches et al., 2019). In agreement, a model of a FHM-3 mutation showed increased SD susceptibility (Jansen et al., 2020). In contrast, loss-of-function mutations in the SCN1A gene are associated with an infantile-onset epilepsy, Dravet syndrome as the reduced excitability in inhibitory neurons disinhibits neural circuitries (Catterall et al., 2010; Yu et al., 2006). The effect of loss-of-function mutation on cortical SD susceptibility is not known, while the high seizure frequency may increase the probability of secondary SD generation even without changing the intrinsic SD threshold.
Some pathogenic gain-of-function mutations in SCN2A/Nav1.2 and SCN8A/Nav1.6 identified in epilepsy patients are associated with increased persistent sodium current in excitatory neurons (Kearney et al., 2001; Veeramah et al., 2012). It is not known whether such mutations also contribute to SD susceptibility in addition to seizure generation. The Nav1.2 channel is interesting in that its conductance is augmented by hypoxia, which may contribute to SD induced under metabolic stress (Hammarström & Gage, 2000; Plant et al., 2016).
In addition to the depolarizing inward conductance, the net charge transfer during the initial slow depolarization is regulated by the counteracting outward current, such as M-type K+ current (Golomb et al., 2006). The M-type current is the non-inactivating voltage-gated K+ current that can persist for seconds or longer. In the mammalian forebrain, the M-current is generated mostly by KCNQ2/Kv7.2 containing K+ channels enriched in the axonal initial segment. Consistent with their role in INaP regulation, genetic impairment or pharmacological modulation of these channels significantly modulates SD susceptibility (Aiba & Noebels, 2021; Y.-J. Wu et al., 2003). These M-current activators reduce ischemic brain injury (Bierbower et al., 2015; Jaeger et al., 2015), and this neuroprotective effect could in part be explained by its SD inhibition property. KCNQ2 mutations are linked to developmental epileptic encephalopathy and contribute to many forms of epilepsy syndromes depending on mutation severity (Lee et al., 2019). The contribution of SD to the neurological deficits in developmental epileptic encephalopathy is not yet known.
In addition to inhibitory roles, K+ conductance such as TEA-sensitive and ATP-sensitive channels may directly contribute to the elevation of extracellular K+ during SD in some contexts (Aitken et al., 1991; Y. Xie et al., 1995). Thus, K+ channels can modulate SD differently depending on their channel gating kinetics and activation modality.
GABAARs provide the major inhibitory synaptic current to forebrain neurons. In hippocampal brain slices, the arrival of SD is preceded by a large GABAAR current, and acute pharmacological inhibition not only increases propagation rate but also occasionally triggers SD without additional stimulation (Aiba & Shuttleworth, 2014; Köhling et al., 2003; Koroleva & Bures, 1983). In contrast, pharmacological augmentation of GABAARs has little effect on the initial GABAA current, likely because the GABAA current is saturated, and did not additionally inhibit SD propagation in acute hippocampus slices (Aiba & Shuttleworth, 2014). As discussed above, FHM3 mutations in SCN1A/Nav1.1 increases SD susceptibility by increased excitability of GABAergic interneurons, and it would increase GABA release (Jansen et al., 2020, p. 3), again suggesting that potentiation of GABA transmission has a limited inhibitory effect on SD. It is notable that GABAAR agonists inhibit SD in the isolated chicken retina (Wang et al., 2015), and thus the efficacy of GABAAR potentiation could vary depending on the brain structure and stimulation.
Similar to SD, activation of GABAARs also has complex effects on seizure susceptibility (Snodgrass, 1992). Diazepam an anesthetic GABAAR agonist, is used to terminate status epilepticus (Falco-Walter & Bleck, 2016), and stiripentol, a GABAAR potentiator, is prescribed for epilepsy associated with GABA hypofunction (Chiron et al., 2000). In contrast, increasing GABA availability or activation of GABABRs paradoxically exacerbates spike-wave seizures by facilitating neuronal synchronization (Liu et al., 1992; Panayiotopoulos et al., 1997), and GABA agonists may trigger seizures in some epilepsy patients (Modica et al., 1990). In contrast to the GABA activators, an impaired GABA system seems to be uniformly pro-convulsive. GABA antagonists such as picrotoxin and pentylenetetrazol are widely used experimental convulsant, and loss-of-function mutations in GABAAR subunits, such as α1, α3, β3, γ2, and δ, are associated with genetic epilepsy (Macdonald et al., 2010, 2012). As discussed earlier, reduced GABAergic inhibitory neuron excitability by SCN1A mutation is a mechanism of developmental epilepsy (Dravet syndrome; Catterall et al., 2010).
Overall, activation/potentiation of GABA signaling has complex effects on seizures and SD depending on the underlying mechanism and context, while impaired GABA inhibition generally contributes to the susceptibility of these excitatory events.
While gap junctions are reportedly required for SD generation in the isolated chicken retina (Nedergaard et al., 1995), they are not required for most forms of SD generated in mammalian brain. Instead, the gap-junction inhibitor, carbenoxolone, or genetic deletion of a gap-junction subunit, connexin-43, facilitated SD propagation in vivo and in vitro (Tamura et al., 2011; Theis et al., 2003). The effect is likely explained by impaired astrocytic K+ spatial buffering. The observation also suggests that the carbenoxolone-sensitive pannexin-1 hemichannels may not be a critical contributor to the initiation/propagation of most forms of SD, although they may contribute to the prolongation of neuronal depolarization (Weilinger et al., 2012). Tonabersat (or SB-220453) is an antimigraine drug that also inhibits SD (Smith et al., 2000). The drug was first characterized as a gap-junction inhibitor, but its SD inhibitory effect is likely mediated by other mechanisms.
Some seizure activities are attenuated by a gap-junction blocker (Gigout et al., 2006; Manjarrez-Marmolejo & Franco-Pérez, 2016), but this effect is likely mediated by inhibition of pannexin-1 hemichannels, rather than gap junctions. These hemichannels contribute to neuronal excitability either by directly depolarizing neuronal membrane potential or via ATP release and activation of purinergic receptors (Aquilino et al., 2020; Dossi et al., 2018; Thompson et al., 2008).
As briefly discussed earlier, the relative contributions of each mechanism can continuously change depending on the metabolism of affected brain tissue (i.e., SD continuum; Dreier & Reiffurth, 2015). In healthy brain tissue, extracellular K+ is effectively managed by astrocytes and does not readily reach the SD threshold level. In fact, an extracellular K+ elevation is extremely small or absent at the wavefront of SD in healthy brain tissue, and its contribution to SD is likely limited. In such a case, SD requires additional contributing mechanisms and thus SD can be inhibited by targeting these contributors. However, under pathological states when the neuronal ATP level is reduced, astrocyte functions are partially impaired, and interstitial space is compressed, extracellular K+ could elevate to the level sufficient to maintain the feedforward depolarization cycles by itself, and thus SD becomes resistant NMDAR antagonists (Jarvis et al., 2001; Lauritzen & Hansen, 1992; Petzold et al., 2005; Y. Xie et al., 1995; E. T. Zhang et al., 1990) and sodium channel blockers (Sugaya et al., 1971). Under such conditions, SD even propagates in the absence of extracellular Ca2+ (Dietz et al., 2009; Zhou et al., 2013). It is generally more challenging to pharmacologically inhibit SD generated in metabolically compromised tissue, since extracellular K+ elevations are more difficult to target. Conversely, some studies suggest partial amelioration of metabolic stress can reveal sensitivity to some inhibitors (Aiba & Noebels, 2021).
Changes in SD contributing mechanisms may also affect propagation velocity and direction. The velocity of SD propagation is increased when contributing mechanisms are enhanced by gene mutations (e.g., FHM mutations) or when the inhibitory mechanism is removed (e.g., Na+K+-ATPase inhibition). SD in a tissue with regionally heterogeneous metabolism often shows a complex SD propagation pattern. In the focal ischemia model, some SD waves propagate in complex patterns, tracing the rim of ischemic penumbra, often confined within the hypoperfused penumbra (Milakara et al., 2017). In a focal KCl-evoked repetitive SD, the first SD spreads centrifugally, but subsequent SDs tend to propagate in irregular patterns (James et al., 1999; Kaufmann et al., 2017).
SD contributing mechanisms could also change when gene expression and network excitability are modified in neurological diseases (e.g., epilepsy, migraine), chronic drug administration, and genetic mutations. SD generated in these conditions may have unique pharmacosensitivities (see Fig. 62–3).
A simplified model of SD contributing mechanisms and pharmacoresistance. SD is generated when contributing mechanisms sum up to exceed the threshold level (dashed line), while it is blocked or attenuated when the net contributions do not reach threshold. (more...)
Astrocytes play critical roles in the regulation of extracellular K+ and glutamate. In order to create the supra-physiological extracellular ionic deregulation during SD, it is necessary to overcome the homeostatic astrocytic function. In this regard, depolarization of astrocytes may be a prerequisite at the leading edge of a propagating SD wave (Sugaya et al., 1975). Compared to SD, depolarization of astrocytes during a seizure is milder and astrocytes likely maintain their homeostatic role.
During SD, astrocytes generate a spreading intercellular Ca2+ wave via gap junctions (Basarsky et al., 1998; Kunkler & Kraig, 1998; Peters et al., 2003). This astrocytic Ca2+ wave is not required for SD propagation itself (Basarsky et al., 1998; Peters et al., 2003), but it may contribute to the cerebral perfusion response (Chuquet et al., 2007; Tóth et al., 2021) and the magnitude of postsynaptic currents following SD (D. C. Wu et al., 2018).
Astrocytic K+ and glutamate clearance are coupled to Na+K+-ATPase activity. Loss-of-function mutations in the principal subunit of glial Na+K+-ATPase, ATP1A2, are identified in cases of familial hemiplegic migraine type-2, FHM-2 (Parker et al., 2021), and result in both impaired glutamate homeostasis and increased SD susceptibility. The activity of astrocytic Na+K+-ATPase depends on ATP production via glycolysis (Pellerin & Magistretti, 1994), and astrocytes reserve a glycogen store to maintain the glucose availability under the metabolic stress (A. M. Brown & Ransom, 2007). Inhibition of glycolysis impairs astrocytic functions and facilitates SD generation (Allen et al., 2005; M. Xie et al., 2008).
The following subsections will review more detailed molecular mechanisms underlying astrocytic K+/glutamate clearance as well as their volume changes relevant to SD and seizures.
Under normal conditions, astrocytes take up excessive extracellular K+ directly by Na+K-ATPases mostly by those composed of the α2β2 complex, which are activated by to extracellular K+ elevation (D’Ambrosio et al., 2002; Larsen et al., 2014). Pharmacological inhibition of Na+K-ATPase by a low concentration of ouabain impairs tissue K+ clearance and prolonged exposure eventually generates an SD (Basarsky et al., 1998; Haglund & Schwartzkroin, 1990).
The inward-rectified K+ channel, Kir4.1, attenuates the local extracellular rise while it contributes little to the acute clearance of the elevated K+ (D’Ambrosio et al., 2002; Larsen et al., 2014). On the other hand, the leak current produced by Kir4.1 is important for maintaining the strongly hyperpolarized membrane potential of astrocytes, and prolonged inhibition of Kir4.1 depolarizes astrocytes by >35 mV (C. B. Ransom & Sontheimer, 1995). In fact, the phenotype of genetic deletion of Kir4.1 in mice is more severe than that expected from acute pharmacological Kir4.1 inhibition. The Kir4.1 KO mouse showed markedly depolarized astrocyte membrane potential, impaired tissue K+ and glutamate clearance, seizure, and premature lethality (Djukic et al., 2007). The pro-epileptic consequences of chronic Kir4.1 deficiency may be relevant to some human epilepsy cases as Kir4.1 expression or its functions is frequently downregulated (Steinhäuser et al., 2012) (Bordey & Sontheimer, 1998; Hinterkeuser et al., 2000). Contrary, some gain-of-function Kir4.1 variants are identified in some epilepsy (Sicca et al., 2011, 2016) while its pathogenic mechanism is unknown.
AQP4 is a water channel expressed in astrocytes and vasculature. Gene deletion of water channel AQP4 impairs K+ clearance and increases the duration of seizures in the mouse (Binder et al., 2006). However, acute pharmacological inhibition of the water channel does not alter K+ clearance (Toft-Bertelsen et al., 2021), and thus the impaired K+ clearance seen in the AQP4 KO mouse is likely due to the chronic developmental effect as seen in the Kir4.1 KO mouse.
The intracellular K+ diffuses to adjacent astrocytes via gap junctions and eventually is damped in the perivascular space (Kofuji & Newman, 2004). Pharmacological inhibition or gene deletion of the gap-junction subunit accelerates the SD propagation rate likely because spatial K+ buffering is impaired (Tamura et al., 2011; Theis et al., 2003). The astrocytic gap-junction network is a dynamic structure and seizure activity disrupts it, which may impair the spatial K+ buffering and facilitate the secondary generation of SD and seizure (Onodera et al., 2021).
Synaptic extracellular glutamate clearance is achieved mainly by the glial glutamate transporters GLT-1 and GLAST (Rothstein et al., 1996). In mouse forebrain, Glt-1 plays a predominant role and gene deletion of Glt-1 results in epilepsy and increased SD susceptibility (Aizawa et al., 2020; Tanaka et al., 1997). Deletion of GLAST is manifested as cerebellar abnormalities in mouse (Watase et al., 1998) and does not affect SD susceptibility in the cerebral cortex (Aizawa et al., 2020); however, mutations in the human GLAST gene (EAAT1) have been associated with cortical seizures and hemiplegic migraine (Jen et al., 2005).
In both Glt-1 and GLAST, extracellular glutamate transport occurs in the net exchange influx of 3 Na+ and efflux of 1 K+ (Vandenberg & Ryan, 2013) and thus is electrogenic (Bergles & Jahr, 1997; Wadiche et al., 1995). Glutamate transporters are physically associated with Na+K+-ATPase (Rose et al., 2009), especially the Na+-sensitive α2β1 Na+K+-ATPase complex, and Na+ influx associated with glutamate intake stimulates ATPase activity, which in turn takes up extracellular K+ (Larsen et al., 2014). Thus, glutamate uptake is coupled to K+ clearance and generally suppresses network excitability.
Intracellular glutamate in astrocytes is converted to inert glutamine by glutamine synthetase (GS). Mutations in GS are associated with epilepsy and premature mortality in humans (Häberle et al., 2005; Sandhu et al., 2021). GS is downregulated in experimentally kindled animals and some epilepsy patients (Eid et al., 2004; Tiffany-Castiglioni et al., 1990). The reduced GS activity may alter network excitability by accumulating intracellular glutamate, which may reduce glutamate uptake rate and also contribute to glial glutamate release upon opening of megachannels, such as volume-regulated anion channels (VRACs; Kimelberg et al., 1990). Glutamine synthesis by GS is also important for intracranial ammonia metabolism, and hyperexcitability of the GS deficient brain may be partly attributable to ammonia toxicity (Rangroo Thrane et al., 2013).
In contrast to glial transporters, the neuronal glutamate transporter EAAC1/EAAT3 plays a limited role in extracellular glutamate clearance, while acute depletion of this transporter results in epilepsy in mice (Rothstein et al., 1996). However, epilepsy phenotype in the EAAC1-deficient mouse is not due to defects in glutamate clearance; rather, it reflects its role in the cysteine uptake and antioxidant glutathione synthesis (Aoyama et al., 2006).
In addition to the uptake mechanisms, astrocytes also affect extracellular K+ and glutamate concentration by changing their cytoplasmic volume. During SD, astrocytes undergo significant swelling (Risher et al., 2009), and the duration of swelling is prolonged when energy substrate is limited (Risher et al., 2012). Depolarization and swelling of astrocytes also appear during seizures (Lux et al., 1986; B. R. Ransom, 1974; Sugaya et al., 1964; Sypert & Ward, 1971; Tønnesen et al., 2018), while these responses are generally milder than during SD. Astrocyte swelling may contribute to SD susceptibility by shrinking the extracellular space, and also by promoting glial glutamate release via VRACs (Kimelberg, 2005; Kimelberg et al., 1990). In support of this, increasing astrocytic volume by hypo-osmotic condition facilitates SD initiation and propagation (Chebabo et al., 1995; Frank et al., 2021; Menyhárt et al., 2020).
Astrocyte swelling occurs following membrane depolarization by elevation of extracellular K+ and glutamate (Walz, 1987; Walz & Mukerji, 1988). In reverse, astrocyte swelling induced by hypo-osmolality shock depolarizes astrocytes (Kimelberg & O’Connor, 1988). This astrocytic voltage-volume relationship has been hypothesized to be mediated by the interplay between K+/Cl− uptake and water influx to maintain cytoplasmic osmolality. However, no studies clearly demonstrated the mechanism. Thus, activity-dependent astrocyte swelling occurs even when potassium uptake mechanisms (i.e., Kir4.1, Na+K+-ATPase) are blocked (Larsen et al., 2014). The major water channel in astrocytes, AQP4, does also not contribute to tissue swelling (Toft-Bertelsen et al., 2021); rather, the water channel is thought to be important for water extrusion (Haj-Yasein et al., 2012; Murphy et al., 2017; Walch et al., 2020). NKCC1 is a Na+, K+, and Cl− transporter and contributes to cytoplasmic osmolality increase during astrocyte swelling in culture and isolated optic nerve (MacVicar et al., 2002; G. Su, Kintner, Flagella, et al., 2002; G. Su, Kintner, & Sun, 2002). However, the NKCC1 expression level is low in adult mammalian brain tissue (Plotkin et al., 1997), and activity-dependent astrocyte swelling has been detected when NKCC1 is pharmacologically inhibited in hippocampal and cortical tissue (Larsen et al., 2014; Steffensen et al., 2015; Walch et al., 2020; Walz & Hinks, 1985). A pH-dependent pathway also contributes to the activity-dependent swelling (Larsen & MacAulay, 2017), but it alone unlikely fully explains the astrocyte swelling response. The underlying molecular mechanisms of astrocyte swelling may vary depending on the pathological context and remain to be elucidated.
Cellular swelling, or cytotoxic edema, is a common feature of acute brain injuries, such as stroke and trauma. The tissue swelling elevates intracranial pressure and, in severe cases, results in cerebral herniation, brainstem compression, and life-threatening cardiovascular dysregulation. In order to prevent malignant consequences, intravenous mannitol infusion is used in acute brain injury. However, a recent study argues that reducing cellular swelling is not uniformly beneficial. Both neurons and astrocytes swell following brain trauma; however, reducing the cytoplasmic volume of neurons elevates their excitability and increases the susceptibility to SD and seizure (P. A. Sawant-Pokam et al., 2020). Identification of astrocyte-type specific swelling mechanisms may minimize such confounds.
Severe ionic deregulation and metabolic stress during SD damage neuronal structures (Kirov et al., 2020; Sword et al., 2013), which could correlate with functional recovery from SD. During SD, neuronal soma transiently swells, even though, unlike astrocytes, neuronal volume is generally well maintained over a wide range of osmolality change (Andrew et al., 2007; Caspi et al., 2009). It is hypothesized that certain water-permeable channels are activated during SD. A study suggests that depolarization-induced neuronal swelling can occur via a Cl− transporter (SLC26A11) dependent mechanism (Rungta et al., 2015), while its contribution to neuronal swelling during SD is not known. The duration of somatic swelling is prolonged in metabolically compromised tissue and results in necrotic cell death.
Dendritic structures undergo complex degenerative changes termed “dendritic beading” characterized by shrinkage and enlargement of dendritic shaft domains (Greenwood et al., 2007). These abnormalities are generally seen in excitotoxicity and involve Na+ and Cl− influx (Greenwood et al., 2007; Steffensen et al., 2015). Interestingly, despite the dramatic morphological changes, the pre- and the postsynaptic terminals remained in contact during SD (Kirov et al., 2020), and ongoing postsynaptic currents during the late SD phase can be detected when the membrane potential is voltage-clamped (Aiba & Shuttleworth, 2012; D. C. Wu et al., 2018). At the subcellular level, mitochondria also undergo severe depolarization and stereotypic structural change, termed “mitochondria-on-a-string” (Kirov et al., 2020; Kislin et al., 2017). Such mitochondrial dysfunction would impact ATP synthesis, oxygen utilization, and neuronal membrane repolarization rate.
These cellular volume and structural changes strongly affect the intrinsic light reflection/scattering properties of the brain tissue. SD is usually associated with biphasic optical signal changes. In healthy tissue, a transient sharp increase in light transparency in association with cellular depolarization phase due to cellular and mitochondrial swelling (Bahar et al., 2000; Zhao et al., 2004) is followed by a slow minutes-long increase, likely reflecting the astrocyte swelling. In a metabolically compromised condition, the secondary transparent phase is replaced by increased light scatter in the dendritic field due to the development of submicrometer dendritic beading. These SD-associated structural changes are exploited for SD detection as the “intrinsic optical signal (IOS)” (Mané & Müller, 2012; Müller & Somjen, 1999). The IOS is detectable in the cerebral cortex in vivo, while the in vivo signal source is complicated due to light absorption by red blood cells. While IOS is useful, these signals contaminate fluorescence imaging signals, especially at the macro- and mesoscopic levels (Valley et al., 2020).
Both dendritic beading and spine loss also occur immediately following a seizure. The severity of structural changes correlates with seizure duration (Guo et al., 2012; Zeng et al., 2007), but they seem milder compared to those detected during SD. On the other hand, seizure-induced structural changes could persist for longer than those induced by an SD in a similar setting and can persist for days or weeks.
The repolarization process is the important determinant of the severity of pathological consequences of SD. This section will discuss the general repolarization mechanism and pathological variants.
The recovery from SD requires glucose and oxygen supply, and reduced plasma glucose and oxygen concentration prolong the duration of depolarization (Hoffmann et al., 2013; Nedergaard & Astrup, 1986; Takano et al., 2007). The utilization of oxygen and glucose for repolarization depends on cerebral perfusion, and systemic hypotension by blood loss prolongs depolarization independently of glucose and oxygen availability (Hoffmann et al., 2012; Sukhotinsky et al., 2010).
At the cellular level, restoration of membrane potential depends on the activity of Na+K+-ATPases. The normalization of Na+ and K+ gradients is also important for the redistribution of other ions, such as Ca2+ and Cl− (Kiedrowski et al., 1994; Zhu et al., 2005). The mammalian brain expresses three isoforms of the Na+K-ATPase catalytic α-subunit in a cell-type-dependent manner. Genetic reduction of each pump gene did not robustly modify the duration of depolarization (Reiffurth et al., 2020), while a disease-linked mutation in ATP1A3 (D801N), a Na+K+-ATPase subunit expressed in both neurons and glia, prolongs the depolarization (Hunanyan et al., 2015). In some forms of SD, NMDARs are activated during the late phase of SD and contribute to the prolonged membrane depolarization (Aiba & Shuttleworth, 2012; Reinhart & Shuttleworth, 2018; D. C. Wu et al., 2018). NMDAR activation under severe metabolic stress also activates intracellular signaling to open the large conductance pannexin1 channel (Thompson et al., 2006; Weilinger et al., 2012;, but see also Madry et al., 2010), and the secondary current may contribute to the pathological prolongation of SD.
The repolarization phase of SD is altered in pathological contexts and, generally, the longer the duration, the higher the chance of neuronal injury and death. The duration of depolarization is prolonged when the local energy substrate supply is insufficient. Prolonged ionic deregulation, such as Ca2+ and Cl−, contribute to structural and functional damages. Prolongation of depolarization by reduced blood supply was demonstrated by Leao (A Leao, 1947) and has been reproduced in subsequent experimental studies. In brain trauma cases, prolongation of DC potential shift correlated with poorer neurological outcomes (Hartings et al., 2011). Leao also reported that the negative DC potential during SD can be temporally neutralized when a cerebral artery was timely occluded (Leao, 1947). While the result is not examined in recent studies, cerebral tissue perfusion may have a more complex effect on the extracellular field potential.
In addition to prolongation, neurons occasionally repolarize from SD with minutes-long recurrent epileptic spiking or seizure activities (Van Harreveld & Stamm, 1953). In some cases, the negative DC potential of SD gradually disappears as it spreads and is detected as a slowly spreading seizure-like activity. These post-SD discharges have been seen in a subset of brain injury patients and in a clinical case associated with behavioral convulsions (Dreier et al., 2012). The incidence of post-SD convulsion is suggested to reflect a pro-epileptic state of the affected brain tissue and may predict epileptogenesis. In migraine cases, these post-SD discharges may underlie the convulsion following migraine (migraine seizure, previously known as “migralepsy”). The post-SD excitation can be reproduced using in vitro preparations by intracellular perfusion of nonselective K+ inhibitor tetraethylammonium (Lin et al., 2017) or pretreatment with GABAAR antagonist bicuculline in vitro (Dreier et al., 2012; Eickhoff et al., 2014). However, this post-SD epileptic activity is not reliably reproduced in vivo, and knowledge of its functional significance is limited.
Together, tissue metabolism and abnormal tissue excitability can modulate the repolarization phase of SD, resulting in prolongation or epileptic discharges. Abnormal recovery from SD seems to correlate with the pathological consequences of SD, such as tissue damage and epileptogenesis (see Fig. 62–4). Since brief waves of SD are relatively benign in healthy brain tissue, pharmacological intervention during conditions of abnormal repolarization would be beneficial for minimizing the deleterious neurological consequences of SD.
Normal and pathological repolarization from SD. Left: Repolarization of SD depends on activities of Na+K+-ATPases and is counteracted by NMDARs in some forms of SD. Middle: In hypoperfused metabolically compromised brain tissue, intracellular ATP synthesis (more...)
SD triggers profound vascular responses, as first documented by Leao (Leo, 1944). He described SD-induced marked vasodilation in the cerebral arteries and microvasculatures which was occasionally followed by hypoperfusion. He also noted that arterial vasodilation was associated with an acceleration of venous blood flow as the venous blood color changes from purple to scarlet, suggesting reduced oxygen extraction. This original observation is generally reproduced in modern studies, while variant responses were also observed under pathological conditions. This section will briefly summarize the cerebral blood flow changes in association with SD and its correlation with migraine pathology. A more systemic review for the neurovascular coupling in SD is available elsewhere (Ayata & Lauritzen, 2015; Dreier, 2011).
The depolarization phase of SD is associated with transient vasodilation and increased cerebral blood flow. In a longer time scale, this SD-induced hyperemia appears overriding on a slow persistent oligemia/hypoperfusion that can last for an hour (Ayata et al., 2004). These vascular responses are species-dependent; the hypoperfusion/oligemia components are more pronounced in the small mouse brain than in rats. The hyperemia phase is also present in human brains, while with a low resolution and slow acquisition rate of MRI, the delayed chronic oligemia (spreading oligemia) is often extrapolated as an SD-associated vascular response (Cao et al., 1999; Hadjikhani et al., 2001; Olesen et al., 1981). This prolonged oligemia could underlie neurological deficits after a migraine attack, such as fatigue and cognitive decline. The molecular mechanism underlying the prolonged hypoperfusion involves cyclooxygenase (COX) signaling pathways (Gariepy et al., 2017; Varga et al., 2016), while other mechanisms also play important roles.
The initial hyperemia response does not necessarily correlate with metabolic demand, but simply reflects the net action of released vasoconstrictive/dilatory factors. In injured hypoperfused brain tissue, these neurovascular couplings are altered due to the reduced vasodilatory factors such as NO as well as profound extracellular K+ surge, which directly depolarizes vascular smooth muscle (Dreier et al., 1998; Hinzman et al., 2014). These changes limit the initial dilatory responses or, in extreme cases, convert the vascular responses to transient vasoconstriction/hypoperfusion (e.g., spreading ischemia; Hinzman et al., 2014; Shin et al., 2006). Abnormal vascular responses can aggravate neurological deficits and contribute to cellular edema associated with necrotic neuronal loss (Dreier et al., 2000). The prolonged vasoconstriction following SD in the ischemic brain creates an enlarged perivascular space, leading to perivascular edema (Mestre et al., 2020). Thus, SD generated in the injured brain significantly contributes to various forms of tissue edema (see Fig. 62–5).
Cerebral blood flow response during SD in anesthetized mouse cortex. An SD was evoked by a KCl application, and blood flow responses were assessed in a distant area (ROI) by imaging the cortical surface. A. Sequence images of cortical surface during (more...)
In contrast to the intracerebral arteries, the meningeal dural arteries and pial veins undergo nearly hour-long vasodilation following an SD (Bolay et al., 2002; Schain et al., 2019). This SD-induced prolonged meningeal vasodilation is suggested to cause extravasation of plasma proteins seen in migraine (Knotkova & Pappagallo, 2007) and contribute to headache via neuroinflammation and trigeminal sensitization. The SD-induced trigeminal activation has been challenged as an experimental artifact due to the direct activation of cerebral meningeal nerves by SD stimulation (e.g., concentrated KCl application). However, a recent study showed that SD induced in the intact skull by the optogenetic method also produced trigeminal sensitization and allodynia (Harriott et al., 2021).
Currently, mechanisms underlying SD-dependent sensitization of the trigeminal nerve system are not fully identified. Calcitonin gene-related peptide (CGRP) is an important mediator of migraine headache. It is released and dilates meningeal vessels following dural stimulation, and clearing the peptide with a monoclonal antibody reduces migraine frequency (Stauffer et al., 2018; Tepper et al., 2017). The CGRP antibody also prevents the SD-induced trigeminal nerve activation/sensitization (Melo-Carrillo et al., 2017), suggesting the CGRP pathway contribute to the SD-dependent migraine mechanism. However, a later study showed that antibody treatment did not inhibit the SD-induced meningeal vasodilation nor plasma extravasation (Schain et al., 2019). It was suggested that the CGRP may be important for the activation of sensory pathways rather than pathological meningeal vascular leakage.
Seizures involve similar vascular responses, characterized by brief hyperemia in association with the discharge activity and subsequent hour-long vasoconstriction and oligemia (Farrell et al., 2016; Gaxiola-Valdez et al., 2017), while the response may vary depending on the seizure type and model. In epilepsy, postictal hypoperfusion may contribute to various clinical postictal deficits (Farrell et al., 2017), and the response has been exploited to map the brain region affected by the seizure (Gaxiola-Valdez et al., 2017). COX2 activity predominantly contributes to the postictal vasoconstriction, and genetic Cox2 deficiency in mice abolishes the vascular response. This strong COX2 dependency of the prolonged oligemia implicate that post-SD/postictal vasoconstriction might be more pronounced in chronic epilepsy because, unlike COX1/3, COX2 expression level is induced by brain hyperexcitation and inflammation (Rawat et al., 2019). The seizure-induced hypoperfusion has not been documented when seizures were acutely evoked in nonepileptic animals (Ma et al., 2013), but seizure-induced hypoxia is reported in other disease conditions such as in glioma and subarachnoid hemorrhage (Montgomery et al., 2020; Winkler et al., 2012). Several studies suggest that cerebral vascular reactivity is altered in epilepsy patients (Diehl et al., 1997; Lv et al., 2018) and potentially increases the risk of vascular disease such as stroke (Shinton et al., 1987).
This section will review the susceptibilities to SD and seizures in the cortical subregions and subcortical structures.
Neocortex is susceptible to SD, but there is a regional differences. SD propagation velocity is increased in the frontal cortex relative to caudal parts (Eiselt et al., 2004; A. A. P. Leao, 1944), while retrosplenial regions are more resistant (Fifkova, 1964; A. A. P. Leao, 1944). On the other hand, KCl application or systemic anoxia stimulation revealed a high SD susceptibility in somatosensory cortex, especially in the mouse barrel cortex (Bogdanov et al., 2016). In rodent brain, the highly developed olfactory system correlates with the formation of the dense neural structure of the barrel columns, which likely contribute to the localized susceptibility, while the visual cortex is a species-specific structure with a high neuronal density and complexity in primate brain (Collins et al., 2010), which may contribute to SD generation in migraine with visual aura. Compared to the dorsal neocortex, the ventral paleocortex is resistant to SD (Fujita et al., 2016). Cortical layer-dependent SD susceptibility has also been identified in rodent brain. In the somatosensory cortex, the upper layers (II–IV) are more susceptible to SD than deeper layers V&VI, and the laminar difference becomes more pronounced when SD is repetitively triggered (Richter & Lehmenkühler, 1993; Joshi & Andrew, 2001; Kaufmann et al., 2017; Nasretdinov et al., 2017; Zakharov et al., 2019; Gniel & Martin, 2013; Juzekaeva et al., 2017; Ichijo & Ochs, 1970). Therefore, intrinsic SD susceptibility is different across cortical regions and layers, and it may bias the expression of SD-dependent neurological deficits. However, the SD incidence is also strongly influenced by the triggering mechanism, presence of tissue injury, and available collateral circulation.
In the gyrencephalic brain, the cortical convolutions also affect SD propagation. The apparent SD propagation rate is faster in the gyrus than in the sulcus, resulting in a complex SD propagation pattern (Schöll et al., 2017). It is generally difficult to predict the exact origin and propagation velocity of an SD based solely on electrophysiological recording data.
In epilepsy, there seems to be no consensus on regional- or layer-dependent neuronal participation in a seizure, as it greatly varies depending on seizure type (Aeed et al., 2020; Zakharov et al., 2019).
SD propagates continuous neural tissue, and anatomical boundaries such as dense white matter often serve as a physical barrier to an SD wave. SD is more prone to occur in the forebrain than in the hindbrain, while the susceptibility is increased during early development. Most of the evidence for the subcortical SD is based exclusively on experimental animal studies, and their clinical significance remains to be elucidated.
The entorhinal cortex and hippocampus form the circuitry critical for memory formation. This structure is rich in excitatory synapses and is frequently involved in the limbic seizures of temporal epilepsy. Both the entorhinal cortex and hippocampus are also susceptible to SD (Andrew et al., 2017); however, there seems to be a anatomical border preventing SD from passing across these structures. Cortically evoked SDs do not readily invade the hippocampus in vivo and in vitro unless conditioned by cannabinoid receptor agonists or induction of long-term potentiation (Martens-Mantai et al., 2014; Tang, Unekawa, Shibata, et al., 2020). In reverse, SD evoked locally within the hippocampus of anesthetized mice, and SDs spontaneously generated in the hippocampus of awake epileptic rats are often confined within the structure (Bahari et al., 2018; Kunkler & Kraig, 2003; Ssentongo et al., 2017).
The striatum is the largest subcortical structure in the mammalian brain. It forms a grey matter nucleus with penetrating white matter bundles and is an important structure to integrate and pass the premotor signals to downstream basal ganglia. One of the unique characteristics of this structure is that the majority of the neurons are GABAergic inhibitory neurons (medium spiny neurons). Thus, unlike other SD susceptible structures, most of the excitatory synapses are formed between long-range axonal projection and local GABAergic neurons, and there is little recurrent excitatory circuitry.
Clinical epilepsy recordings suggest that seizures rarely spread in the striatum (Aupy et al., 2019; Kuba et al., 2003). However, this structure is highly susceptible to SD in the rodent brain (Andrew et al., 2017; Umegaki et al., 2005), and interestingly SD can spread relatively easily across the striatum and cerebral cortex. A KCl-evoked SD in the striatum can propagate into the cerebral cortex (Vinogradova et al., 1991), and in reverse, cortically evoked SD invades the striatum (Eikermann-Haerter et al., 2011). The major cortico-striatal connection is a long-range white matter tracts, and it is unlikely that SD passes through them. Instead, cortico-striatal SD propagation may be made through grey matter bridges, caudally via the amygdala, or rostrally via the claustrum and nucleus accumbens. These subcortical grey matter paths may not be well conserved in the human brain, and the clinical significance of cortico-striatal SD remains to be validated.
Striatum and hippocampal invasion of cortically evoked SD is enhanced in the mouse carrying the FHM1-linked PQ-type calcium channel and likely FHM-2 linked ATP1A2 mutation (Cain et al., 2017; Eikermann-Haerter et al., 2011; Tang, Unekawa, Shibata, et al., 2020). The subcortical invasion of cortically evoked SD in FHM mice was inhibited by pregabalin, an analog of gabapentin that inhibits SD initiation and propagation (Cain et al., 2017). SD susceptibility may generally correlate with subcortical SD invasion.
Several studies suggest the presence of subcortical structure-specific neurochemical SD mechanisms; SD in the striatum is uniquely dependent on D1 receptor activity (De Lure et al., 2019), while in nucleus accumbens, D2 antagonists exacerbate the depolarization (Hobbs et al., 2017). These tissue-specific mechanisms raise the possibility that diseases associated with specific neuromodulatory pathways may affect regional subcortical SD susceptibility.
The cerebellum is involved in fine motor control, and injury is associated with dystonia, ataxia, speech impairment, and cognitive decline. Clinical studies are revealing its roles in the cognitive function (Schmahmann, 2019), especially the posterior lobe, and dysfunction may contribute to some psychiatric disorders. The cerebellar cortex is composed of three layers. The superficial molecular layer contains high-density glutamatergic excitatory synapses between granule cell axons (parallel fibers) and Purkinje cells. The second layer is the Purkinje layer, and third is the inner granular cell layer, whose main function is to relay cerebellar input (mossy fiber) to Purkinje cells. Unlike the cerebral cortex, the cerebellar cortex contains very few recurrent excitatory synapses and is less seizure-prone.
The cerebellum is vulnerable to ischemic injury; however, this structure is relatively resistant to SD (Oliveira-Ferreira et al., 2020) and is supplied with vascular structures distinct from the cerebral cortex. A recent study suggests a deficiency in PRRT2, a gene linked to developmental epilepsy, can significantly increase SD susceptibility and contribute to transient dyskinesia (B. Lu et al., 2021). It is not known whether cerebellar SD can spontaneously occur and contribute to paroxysmal events.
The brainstem is generally resistant to SD in adult animals (Andrew et al., 2017; Brisson et al., 2014); however, the SD threshold is significantly lowered in young developing animals (Funke et al., 2009; Richter et al., 2003). The developmental high SD susceptibility is likely due to higher neuronal density, synapse structures, and low white matter content (Vincent & Tell, 1999). When combined with gene mutations associated with higher SD susceptibility, the structure becomes vulnerable to lethal brainstem depolarization after seizures (Aiba et al., 2016; Aiba & Noebels, 2015; Jansen et al., 2019; Kron et al., 2011). In experimental studies, SD appeared in the brainstem secondary to a cortically induced seizure, and the SD onset was associated with cardiorespiratory shutdown, which may be relevant to the sudden unexpected death in epilepsy (SUDEP) (see Chapter 63, this volume). These studies are solely based on rodents, and the significance of brainstem SD in human SUDEP awaits validation. It is not known whether other structures that are resistant to SD in the healthy adult are also vulnerable to SD at certain ages or genetic conditions.
The second part of this chapter will discuss SD and seizure co-generation in acute, subacute, and chronic disease conditions, with a major focus on acute brain injury, migraine, and epilepsy.
Acute brain hypoxia/ischemia is a well-known trigger of SD, and these SD events contribute to both initial lesion development and secondary infarct growth (Dohmen et al., 2008; Fabricius et al., 2008; Hartings et al., 2003). Correlatively, the reduced cerebral perfusion pressures increase the susceptibility to SD in brain trauma patients (Hartings et al., 2009).
In brain slice models, removal of oxygen alone in an interface chamber, or in combination with glucose omitted from the perfusate in the submerged condition (e.g., oxygen-glucose deprivation [OGD]), are commonly used triggers of SD (Croning & Haddad, 1998). In the past literature, these forms of SDs are referred to as anoxic depolarization, hypoxic depolarization, or ischemic depolarization.
SD initiation by cellular depolarization due to the intracellular ATP depletion seems intuitive; however, the underlying neurophysiological mechanisms are complicated. In general, acute metabolic stress rapidly suppresses neuronal excitability by opening K+ channels which leads to membrane hyperpolarization and reduced membrane resistance (Leblond & Krnjevic, 1989). Activity-dependent synaptic currents are also suppressed by adenosine receptor activation (Gervitz et al., 2001), while this increases activity-independent spontaneous excitatory and inhibitory synaptic activities (Fleidervish et al., 2001), and a gradual extracellular K+ elevation and tissue swelling (Croning & Haddad, 1998; Rice & Nicholson, 1991; Vorísek & Syková, 1997). These complex responses persist until neurons begin to discharge and then fully depolarize upon arrival of SD (Allen et al., 2004; Müller & Somjen, 2000). It is noted that most of these observations are made from sites away from the SD initiation site; the course of neuronal excitability changes at the precise site of origin during the emergence of SD is not fully characterized.
In addition to a metabolic effect, hypoxia increases SD susceptibility by modulating ion conductance by shifting the extracellular redox potential toward “reduced.” The redox-dependent mechanism involves modulation of the BK channel (Hepp et al., 2005), persistent sodium current (Hammarström & Gage, 2000), and NMDARs (Aiba & Shuttleworth, 2013; Aizenman et al., 1990).
In clinical cases, the seizure is a comorbidity of stroke and can occur in association with SD (Fabricius et al., 2008) and the post-stroke seizure may predict the risk of future seizure incidence and epileptogenesis (Camilo & Goldstein, 2004). Cerebral ischemia-induced seizures are seen after experimental stroke (X.-C. M. Lu et al., 2009; Song et al., 2018; Williams et al., 2008). These experimental ischemia-induced electrographic seizures are typically nonconvulsive and behaviorally silent.
In addition to stroke, systemic hypoxia/hypoxemia also triggers seizures. The infant brain is especially susceptible to hypoxic seizures, which increases the risk of epilepsy in later life (Vasudevan & Levene, 2013). Experimentally, seizures can be triggered by exposure to hypoxic gas (3%–4% ambient oxygen) in immature but not in adult rats (F. E. Jensen et al., 1991; Rakhade et al., 2011). Hypoxia rarely triggers seizures in acutely prepared brain slices unless extracellular K+ concentration is artificially elevated (Kawasaki et al., 1990). In contrast to SD (Hartings et al., 2009), hypoxia/hypotension is not a common prodrome of focal epilepsy. Arterial blood pressure is typically increased and oxygen saturation is unaltered before focal seizure onset (Hampel et al., 2016).
Glucose availability in the brain modulates SD susceptibility. In ischemia studies, SD onset is delayed by hyperglycemia, while hypoglycemic conditions increase SD susceptibility (Hansen, 1978; Hoffmann et al., 2013). These effects seem to be independent of a systemic effect, as acute inhibition of glycolysis hastens OGD-induced SD onset in brain slices (Allen et al., 2005). Severe hypoglycemia alone may also induce SD in vivo, and the onset is associated with coma in rats (Siesjö & Deshpande, 1987; E. T. Zhang et al., 1990).
Seizures are rare in acute hypoglycemia (Imad et al., 2015); however, chronic intracranial glycemic imbalance is linked to epilepsy. Gene mutations in the glucose transporter responsible for glucose transport across the blood–brain barrier (GLUT1 deficiency syndrome) are commonly associated with epilepsy. Some of these patients also experience hemiplegic migraine (Almuqbil et al., 2018; Gburek-Augustat et al., 2020; Mohammad et al., 2014; Pellegrin et al., 2017), and the condition may also be linked to SD susceptibility.
Brain trauma is frequently associated with SD. In the severe cortical contusion model, injury is associated with a large increase in K+ and glutamate levels in the surrounding tissue, a condition suggestive of SD (Katayama et al., 1990; Maeda et al., 1998). The extracellular K+/glutamate surge can be attenuated by a nonselective Ca2+ channel inhibition by cobalt or the glutamate receptor antagonist kynurenic acid, but not by the sodium channel inhibitor TTX. These pharmacological properties suggest presynaptic glutamate release and postsynaptic glutamate receptor activation drives SD by elevating extracellular K+. Most brain trauma models involve hemorrhage and direct brain tissue damage, and SD generation in these models is therefore multifactorial. SD also arises from reduced tissue perfusion and vascular damages secondary to the initial trauma injury.
In the rodent brain, SD can even be triggered by a pinprick without causing overt damage, suggesting minor mechanical stress alone is capable of generating SD. This pinprick SD is inhibited by postsynaptic glutamate receptor antagonists (Akerman & Goadsby, 2005; Kaube & Goadsby, 1994); however, it is resistant to inhibition of voltage-gated P/Q- and N-type calcium channels (Akerman et al., 2008; Richter et al., 2002). Pinprick SD is also inhibited by acid-sensitive ion channel (ASIC1) inhibitors (Holland et al., 2012). These pharmacological profiles suggest a unique initiation mechanism in mechanically induced SD. It is notable that these pharmacological studies were conducted in different laboratories using different preparations and often lacked controls, and therefore the underlying mechanisms require further examination.
Increased brain temperature generally increases tissue excitability and may contribute to seizure (i.e., febrile seizure; Dubé et al., 2007). Infants and young children are more susceptible to febrile seizures than adults (Verity et al., 1985). Febrile seizure induction likely involves changes in ion channel kinetics, while systemic effects such as tissue alkalization due to hyperventilation also play an important role (Schuchmann et al., 2006). At the circuit level, heat inactivation of the inhibitory system is suggested as an important seizure initiation mechanism (Kang et al., 2006; Tran et al., 2020). In fact, interneuronal hypofunction lowers the thermal threshold for the febrile seizure (e.g., Nav1.1/Scn1a loss-of-function mutation in Dravet syndrome; Oakley et al., 2009).
Brain hyperthermia is also associated with higher SD susceptibility. SD frequency in intracerebral hemorrhage patients is increased when the brain temperature is higher, and SD is more often detected following the elevation of body temperature (Hartings et al., 2009; Schiefecker et al., 2018). In in vitro preparations, hyperthermia alone can induce SD in acute brain slices prepared from the immature rat brain when tested in an interface chamber (J. Wu et al., 2003; J. Wu & Fisher, 2000). Such hyperthermia SD can be generated in the presence of TTX, suggesting febrile-SD arises independently of neuronal discharges. The intracellular ATP level was unchanged; thus, metabolic stress is not a critical contributor to SD in this model. SD is also sensitive to reduced temperature, and most forms of SD fail to generate in vitro when experiments are conducted at room temperature (Morris et al., 1991).
In seizure-naïve brains, seizure and SD can co-generate in temporally and spatially associated manners, forming preictal-SD and postictal-SD complexes. The pre-ictal SD, an SD followed by seizure activity, is a potential biomarker of epileptogenesis. The postictal SD, an SD immediately following a seizure discharge, has been detected in numbers of epilepsy studies in vitro and in vivo. While most spontaneous seizures terminate without evidence of SD, postictal SD transiently ceases ongoing seizure activities and creates a brief interictal period (Heuser et al., 2018; Tamim et al., 2021). As discussed earlier, extracellular K+ concentration rises to 10–14 mM during seizure activity, a concentration that is close to a minimal exogenous K+ concentration to trigger an SD in vitro and in vivo (Bogdanov et al., 2016). The severe seizure also accompanies prolonged tissue hypoxia. Overall, the postictal state with elevated extracellular K+ and metabolic stress could create a time window with increased SD susceptibility. Nonetheless, not all seizures are followed by SD, and most seizures self-terminate without SD.
Several studies suggested that SD waves tend to avoid or fade around tissue regions with concurrent seizure activity (Koroleva & Bures, 1979, 1980). Counterintuitively, a study even showed that higher extracellular K+ concentration correlated with stronger SD inhibition (Ueda & Bures, 1977). While the detailed mechanism is not known, seizure foci are not a uniformly favorable condition for SD propagation, and this anti-SD property may set a local threshold for the emergence of SD following a seizure.
Postictal SD is sometimes considered as a mechanism for abrupt suppression of EEG activity, or postictal generalized EEG suppression (PGES); however, in many cases this is not correct. Unlike SD, which is detected as a spreading wave of EEG attenuation, PGES occurs simultaneously across large areas of the brain, and the suppression is usually more severe. Neurobehavioral correlates between PGES and SD-induced EEG depression may also differ. PGES is often associated with reduced cognition, while cortical SD does not always produce a decreased state of consciousness. PGES is thought to involve changes in the output of the brainstem reticular activating system, and because of its close functional connectivity with central autonomic circuitries, PGES can be associated with life-threatening cardiorespiratory depression and sudden unexpected death in epilepsy (Peng et al., 2017). In contrast, acute neurological deficits directly attributable to SD seem to correlate with the functions of the affected brain structure, along with secondary activation of the trigeminovascular system due to the meningeal inflammation (Moskowitz et al., 1993).
Both clinical and basic studies strongly suggest that the majority of migraines with aura are likely attributable to SD originating in the occipital lobe. The visual aura of migraine is typically a slowly migrating blurred spot in the visual field, and MRI studies suggest this is a direct effect of the SD wave propagating in the visual cortex (Cao et al., 1999; Hadjikhani et al., 2001; Olesen et al., 1981). The strongest clinical evidence comes from the MRI imaging of blood flow during migraine attack which demonstrated an oligemia spreading from the occipital lobe (Hadjikhani et al., 2001). Experimental animal studies provided mechanistic insights into the migrainous symptoms caused by SD. As discussed earlier, SD triggers prolonged meningeal vasodilation and extravasation (Bolay et al., 2002; Schain et al., 2019), which may contribute to activation of the trigeminal nerve system. Correlatively, cortical SD is sufficient to create migraine-related such as allodynia, photophobia, and anxiety (Harriott et al., 2021; Tang, Unekawa, Kitagawa, et al., 2020). Together, both clinical and experimental studies suggest that cortical SD contributes to the pathology of migraine with aura.
Because of the link between cortical SD and migraine with aura, the peri-ictal migraine headache is sometimes assumed to be due to cortical SD generated in association with a seizure; however, the relationship is unlikely to be that simple. Migraine with aura represents ~25%–30% of migraine cases, and the contribution of SD to the other class of migraine headaches is not known. In addition, while most peri-ictal migraines involve visual aura, most visual auras are distinct from the stereotypic one reported by migraine with aura patients; the visual aura in epilepsy is shorter in duration and appears in a hemifield, independently of nausea and photophobia commonly present in migraine with aura (Hartl et al., 2017). As discussed above, a seizure by itself generates SD-like vascular responses, which may be sufficient to trigger a headache; thus, the occurrence of peri-ictal migraine may not necessarily implicate an SD.
In contrast to the high prevalence of migraine in epilepsy, seizures are a rare comorbidity of simple migraine, although migraine patients as a group do have an increased risk of seizure (Andermann, 1987). Even in the SD-linked disease, familial hemiplegic migraine, the overall prevalence of seizures is less than 40%, and most of the seizures do not temporally associate with a migraine attack; rather, they arise secondary to edema or fever (Prontera et al., 2018). Therefore, genetic conditions rendering a high susceptibility to SD may not generally contribute to increased seizure susceptibility.
Epilepsy has long been considered an SD-related pathology; however, clinical evidence of spontaneous SD incidence in general epilepsy patients is quite limited. One of the confounds is that the slow DC potential shift of SD can be severely attenuated by the skull, and scalp surface EEG recording is always contaminated by electrodermal activity. However, surgical treatment of epilepsy is increasingly popular, and as chronic intracranial ECoG recordings become more available, there are increased opportunities to explore SD incidence in epilepsy patients using a DC amplifier or AC amplifier with appropriate settings. In fact, under these conditions, a clinical study has reported SD-like electrophysiological events in epilepsy patients (Bastany et al., 2020). However, It is noted that SD duration and amplitudes reported in the study were quite small, and similar millivolt DC shifts have been detected and reported as “infra-slow activity” (Rodin et al., 2014). Intraoperative recording during epilepsy resection surgery is another opportunity to gain access to ECoG recording. However, recording duration is generally short, and the use of anesthesia and perioperative anticonvulsant medication could confound the spontaneous SD generation relevant to epilepsy (Santos et al., 2019).
The absence of strong clinical evidence could also suggest that SD might be rare in most epilepsy cases. In fact, some epilepsy or pro-epileptic conditions could be associated with increased SD thresholds. NMDAR is one of the key contributors of SD; however, although rare in number, loss-of-function mutations and autoimmunity to NMDAR subunits are associated with epilepsy (Florance-Ryan & Dalmau, 2010; Gable et al., 2012; Xu & Luo, 2018). Gain-of-function mutations in P/Q-type Ca2+ channels are linked to familial hemiplegic migraine, but a loss-of-function mutation in the P/Q type channel elevates the SD threshold and contributes to absence epilepsy (Ayata et al., 2000; Miao et al., 2020). Nav1.1 gain-of-function mutations are linked to familial hemiplegic migraine type-3; however, loss-of-function mutations in this channel are linked to infantile-onset epilepsy, Dravet syndrome. In addition to these rare genetic cases, chronically epileptic brain tissues from rodent and human patients have increased K+ threshold for SD when tested in vitro (Köhling et al., 2003; Maslarova et al., 2011), despite the fact that these brain tissues are more susceptible to seizure (King et al., 1985; Oliver et al., 1980; Stringer & Lothman, 1988). Thus, even in an apparently hyperexcitable epileptic brain, SD susceptibility might instead be reduced. It will be important to individually document seizure type, history, and mechanism when characterizing SD incidence in clinical epilepsy.
Both clinical and basic studies report that SD and seizures are expressed in brains with altered network excitability, supporting the idea that SD is a comorbidity of epilepsy. However, most seizure-SD interactions have been studied in the seizure-naïve brain, and preclinical evidence for SD generation in the clinical epilepsy-relevant condition of chronic seizures is quite limited. In order to gain insight into SD in epilepsy, we recently characterized the spontaneous seizure and SD phenotype of two genetic epilepsy mouse models, Kv1.1-KO and Emx-cre:Kcnq2flox/flox (Kcnq2-cKO) mice (Aiba & Noebels, 2021). This study identified distinct genotype-dependent SD generation patterns; the Kcnq2-cKO mouse had a unique stereotypic SD generation pattern, characterized by a bilateral symmetrical propagation, strong seizure-SD coupling, and frequent generation during the dark phase. In contrast, most seizures in the Kv1.1 KO mouse were generated without SD, and even when SD was present, it was generated unilaterally and with seconds of latency after seizure termination. These initial results from genetic mouse epilepsy models again emphasize a variability of the SD phenotype depending on the underlying molecular defect. These preclinical data are essential to understanding the mechanism, predicting the risk, and selecting the appropriate treatment of SD in epilepsy patients.
This work was supported by American Heart Association career development grant 19CDA34660056 (I.A.), Curtis Hankamer Basic Research Fund at Baylor College of Medicine (I.A.), American Epilepsy Society Junior Investigator Award (I.A.), and NIH Center for SUDEP Research (NS090340 and NS29709, Jeffrey L. Noebels)
The author declares no relevant conflicts.
This is an open access publication, available online and distributed under the terms of a Creative Commons Attribution-Non Commercial-No Derivatives 4.0 International licence (CC BY-NC-ND 4.0), a copy of which is available at https://creativecommons.org/licenses/by-nc-nd/4.0/. Subject to this license, all rights are reserved.
Your browsing activity is empty.
Activity recording is turned off.
See more...
