|Home | About | Journals | Submit | Contact Us | Français|
Abnormally enhanced glutamatergic excitation is commonly believed to mark the onset of a focal seizure. This notion, however, is not supported by firm evidence, and it will be challenged here. A general reduction of unit firing has been indeed observed in association with low-voltage fast activity at the onset of seizures recorded during presurgical intracranial monitoring in patients with focal, drug-resistant epilepsies. Moreover, focal seizures in animal models start with increased γ-aminobutyric acid (GABA)ergic interneuronal activity that silences principal cells. In vitro studies have shown that synchronous activation of GABAA receptors occurs at seizure onset and causes sizeable elevations in extracellular potassium, thus facilitating neuronal recruitment and seizure progression. A paradoxical involvement of GABAergic networks is required for the initiation of focal seizures characterized by low-voltage fast activity, which represents the most common seizure-onset pattern in focal epilepsies.
The understanding of the transition from the interictal condition into seizure generation (ictogenesis) is crucial in epileptology. Neurology textbooks as well as most reviews and original papers regard focal seizures as the expression of excessive synchronization of excitatory neuronal networks. This statement rests on the observation of large amplitude electroencephalography (EEG) transients during seizure activity both in patients presenting with focal epilepsy and in animal models of epilepsy. However, the presence of large, synchronous (and presumably excitatory) EEG events during a seizure does not necessarily imply that analogous mechanisms are responsible for its initiation. The events occurring at seizure onset may, therefore, be different from those involved in maintaining seizure activity. Moreover, the network and cellular mechanisms responsible for seizure generation are likely diverse in different forms of focal epilepsies.
Large amplitude EEG transients represent the most evident pattern associated with interictal activity in several forms of focal epilepsy. In patients as well as in experimental models of focal epilepsy, interictal spikes (i.e., short-lasting, large-amplitude EEG events occurring between seizures) correlate with brief increases in neuronal firing and are followed by a profound depression in excitability associated with a slow-wave component.1 Based on in vivo intracellular recordings obtained from acute models of epileptiform synchronization, it was proposed >40 years ago that the transition from interictal to ictal activity results from weakening of the postspike “inhibitory” processes leading to an uncontrolled increase in excitation.2 According to this hypothesis, seizures occur when excitatory glutamatergic transmission overcomes inhibition, thus transforming the short-lasting interictal event into a prolonged, large-amplitude epileptiform discharge. However, this scenario has been never confirmed. Rather, clinical evidence suggests that focal seizure onset is often characterized by decreased amplitude and “flattening” of the scalp EEG that correlates with low amplitude activity recorded with intracranial electrodes routinely utilized to identify the boundaries of the epileptogenic zone in patients with drug-resistant epilepsy candidate for epilepsy surgery.
Intracranial human data3 and experimental evidence in animal models in vivo4,5 and in vitro6,7 have demonstrated that the most common EEG-onset pattern observed in focal seizures/epilepsies is characterized by low voltage fast activity (preceded by ictal spikes or not), whereas large amplitude discharges represent the established seizure core. As illustrated in Figure 1, human intracerebral recordings have demonstrated that seizure onset correlates with (1) waning of background activity, (2) the appearance of low-voltage fast activity in the beta-gamma range (30–100 Hz), and (3) a very slow field potential shift.8–11 Such low-voltage fast focal-onset pattern gradually evolves within seconds into large-amplitude EEG discharges that are unmistakably recognizable as seizure activity. Of interest, the low-voltage fast seizure onset observed both in temporal lobe structures and in neocortical regions can be heralded by one or more large-amplitude synchronous potentials, 8,11,12 designated as preictal spikes (see subsequent text).
In this review we challenge the common belief that enhancement of glutamatergic transmission underlies the transition from interictal to ictal activity in focal seizures with low-voltage fast onset. We review clinical and experimental evidence that underscores the paradoxical and unexpected role played by γ-aminobutyric acid (GABA)ergic networks in seizure initiation. The preservation of GABAergic neurons in humans and in animal models of focal epilepsies, albeit controversial, is supported by numerous reports.8,13 First, we address the results obtained in vivo by analyzing single unit activity in patients with epilepsy and in chronic animal models. Then, we discuss the role played by GABAA receptor signaling in seizure initiation. Finally, we propose a mechanistic model of ictogenesis that highlights the concept that synchronization of interneuron networks leads to increases in extracellular potassium that support seizure progression. Our hypothesis is specifically proposed for focal seizures that initiate with low-voltage fast activity and represent the most common seizure-onset pattern in focal epilepsies.8,14
Early work performed in patients with drug-resistant epilepsy during presurgical EEG monitoring casted doubt on the experimental view that transition from interictal to ictal activity could result from increased excitation, as proposed by Ayala et al.2 Specifically, it reported a widespread reduction of neuronal firing at seizure onset, possibly due to inhibition of principal cells.15 More recently, intracranial presurgical recordings with microelectrode arrays in patients with focal epilepsy have confirmed a reduction of firing in a large population of single neurons at seizure onset16 (Fig. 2A). These data suggest that the low-voltage fast activity seen in the EEG at the beginning of focal seizures correlates with an arrest of principal cell firing possibly reflecting enhanced inhibition in the epileptogenic region. In addition, Schevon et al.17 have demonstrated irregular and relatively low level of unit firing at the onset of ictal activity in patients with focal epilepsy; this phenomenon was interpreted by these authors as an inhibitory restraint that prevents the spread of the ictal discharge (see subsequent text).
The decreased firing of principal, glutamatergic neurons along with the enhanced activity of presumptive inhibitory cells has been documented in vivo with single-unit recordings performed in the hippocampus of pilocarpine-treated epileptic animals, which represent an established model of temporal lobe epilepsy.18–20 These studies have shown that the initial 5–10 s of a seizure correlate with a virtual cease of principal neuron activity along with interneuron firing enhancement (Fig. 2Ba); such a pattern could be identified in >150 principal cells and about 50 interneurons recorded during several focal seizures18 (Fig. 2Bb). Of interest, desynchronization of principal neuron firing21 along with an increased activity of putative interneurons22 was also identified at the transition from interictal to ictal activity in acute models of focal ictogenesis in vivo.
Overall these studies indicate that principal neurons are not hyperactive at seizure onset, presumably as the result of the increased firing of interneurons. However, a causal relation between increased interneuron activity and seizure initiation has not been identified in vivo. In the following section we address studies that have revealed the presence of synchronous GABAergic events at the start of focal seizure activity in vitro, and have provided evidence for a mechanism implicating GABAA receptor signaling in inter-ictal to ictal transition.
As shown in Figure 3A, ictal discharges recorded in vitro from principal neurons in several limbic structures during application of the potassium channel blocker 4-aminopyridine start with an intracellular depolarization that is associated with few if any action potentials.23 Paradoxically, this initial depolarizing component—which is mirrored by field activity—becomes hyperpolarizing when the neuron membrane potential is brought to values less negative than −60 mV with steady injection of depolarizing current (right trace in Fig. 3Ab). Therefore, in this in vitro model of epileptiform synchronization, ictal-discharge onset is associated with a powerful GABAergic event. In line with these data, Ziburkus et al.24 have reported increased activity of interneurons at the onset of ictal discharges induced in vitro by 4-aminopyridine. Moreover, pharmacologic manipulations that antagonize GABAA receptors or decrease presynaptic release of GABA from interneurons, abolish this initial synchronous event and prevent ictal discharge occurrence.6 In vitro data supporting the participation of GABAergic inhibition in the initiation of seizure activity are not specific to the 4-aminopyridine model; similar findings have also been obtained in other models of focal seizures induced in brain slices by high-frequency tetanic stimulation,25–28 in the isolated immature hippocampus by application of low magnesium medium,29,30 and in the in vitro isolated guinea pig whole brain following proconvulsive drug treatment.31–33 As illustrated in Figure 3B, fast beta-gamma activity observed at the start of an ictal discharge in the entorhinal cortex of the isolated guinea pig brain is paralleled by intracellular hyperpolarizing potentials that become smaller in amplitude as seizure progresses, suggesting a dynamic positive shift of the GABAA-mediated inhibitory post-synaptic potential (IPSP) reversal potential.29,34
It should be remarked that fast-onset activity in seizures studied in vitro is usually preceded by one or more preictal spikes that correlated with bursting activity of GABAergic interneurons (Fig. 3C, right traces) and with inhibitory potentials in principal neurons (Fig. 3C, left traces31). Synchronous GABAergic potentials during the interictal period and just ahead of a focal seizure have been observed in brain slices obtained from different cortical areas and in the isolated guinea pig brain preparation.23,32–34 As mentioned earlier, sequences of large amplitude spikes were observed at seizure onset in both humans and animal models in vivo.11,35–37 Such preictal spikes are different from inter-ictal spikes and should be considered as proper ictal events, since they occur exclusively at seizure onset. Of interest, seizure onset in human partial epilepsy has been hypothesized to be due to a reduction of GABAergic dendritic inhibition releasing an enhancement of somatic inhibitory interneurons generating fast IPSPs on pyramidal cells.38
The slow negative shift of the local field potential observed in association with preictal spikes and the ensuing low-voltage fast activity (lower extracellular traces in Fig. 3A–C) correlates with elevations in extracellular potassium (Fig. 3B)31 that are known to accompany synchronous neuronal activity.39–42 However, and even more important in the context of our hypothesis stating that seizure onset rests on “excessive GABAergic signaling,” increases in extracellular potassium are induced by activation of GABAA receptors following application of GABA39 or sustained interneuron activity, such as that occurring at seizure onset in rodent brain slices and in slices of human dysplastic cortex maintained in vitro.23,25,43–46
An explanation for these GABAA receptor–dependent elevations in extracellular potassium rests on the intracellular chloride accumulation that is known to occur in neurons once postsynaptic GABAA receptors are activated; this process increases the activity of the potassium-chloride cotransporter (KCC2), thus causing the efflux in the extracellular space of both potassium and chloride ions.47 In addition, GABAA receptor–dependent elevations in extracellular potassium can also be contributed by network-driven, bicarbonate currents.45,46 These mechanisms of extracellular potassium elevation mediated by GABA release are expected to involve an extensive population of mutually interconnected and electrically coupled interneurons that innervate a large number of principal cells.
Increased extracellular potassium concentration is well known to cause hyperexcitability and epileptiform synchronization48 by (1) depolarizing neuronal membranes, (2) promoting the generation of small amplitude, possibly ectopic spikes49 (arrows in Fig. 3A, B), (3) releasing the resonance of local networks to generate oscillatory patterns in the beta-gamma range,50 and (4) causing a positive shift of the membrane reversal of GABAA receptor–mediated inhibitory postsynaptic potential that weakens inhibition and releases intrinsic excitatory mechanisms.11
The pivotal role played by interneurons in ictal discharge initiation has recently been confirmed with optogenetic techniques in the entorhinal cortex during 4-aminopyridine treatment.51,52 In these studies, optogenetic stimulation of parvalbumin- or somatostatin-positive interneurons initiated ictal events similar to those occurring spontaneously (Fig. 3D); when the onset of an optically induced ictal discharge was expanded, the typical pattern characterized by one or two interictal spikes leading to the fast beta-gamma ictal activity could be identified (arrowheads in Fig. 3D). We should also mention that optogenetic studies have demonstrated that activation of GABAergic cells contributes to seizure maintenance53 and can play a more conventional inhibitory role by contributing to the termination of ictal discharges.54,55
Taking into consideration the clinical and experimental evidence reviewed here, we propose a model of focal ictogenesis whereby GABAergic network activity initiates (and presumably maintains) seizures through the activation of GABAA receptors that are postsynaptically located mainly on principal glutamatergic cells. As mentioned earlier, we propose that “preictal” spikes reflect interneuronal network synchronization at seizure onset and thus belong to the ictal condition.
As summarized in Figure 4, interictal spikes are largely contributed by short-lasting synchronous discharges of action potentials generated by principal cells. Such activity is restrained by several mechanisms, including recurrent synaptic inhibition, that limit excitation both spatially and temporally, and hinder seizure occurrence. EEG spikes subsidized by interneuronal firing herald the period that immediately precedes seizure onset. The intense and possibly synchronous discharge of action potentials generated by interneurons during such spikes leads to a large release of GABA that inhibits and silences principal cells via activation of postsynaptic GABAA receptors. The cotransport of potassium and chloride out of principal neurons consequent to the activation of GABAA receptors will cause sizeable increases in extracellular potassium that depolarizes all neurons and contributes to the emergence of resonant low-voltage fast-network activity. Further increases in potassium also facilitate recruitment of principal cells, leading to overt seizure discharge, which is characterized by progressive, excessive, and simultaneous coactivation of glutamatergic and GABAergic networks.
Several researchers have confirmed the occurrence of GABAergic synchronization at seizure onset, but have provided different interpretations of ictogenesis. For instance, Trevelyan and colleagues17,56 proposed that GABAergic activation represents an inhibitory restraint occurring ahead of the ictal excitatory wave and thus opposes the generation of epileptiform activity. This interpretation, however, rests on the nonverified assumption that excitatory-based ictal discharges have already started somewhere else (as in the model of focal seizures induced by local N-methyl-D-aspartate (NMDA) applications57) and on the presumption that GABA receptor signaling has a compensatory (reparative) function rather than an active role in seizure generation as reviewed earlier.
A possible interpretation of the role played by GABAergic synchronization at seizure onset is based on data obtained from immature/juvenile rodent brain tissue; specifically, it proposes that excessive interneuron activity during epileptiform discharges induces a shift of the GABAA receptor–mediated chloride reversal potential, thus transforming GABAergic signaling from inhibitory to excitatory. 58–60 However, neuronal firing during depolarizing GABA shifts was never demonstrated in adult animal models of seizures/epilepsy.
The clinical and experimental data reviewed herein suggest that the role of GABAergic networks in focal seizures is more complex than what was assumed in the past. The general belief that a reduction of GABAergic activity is proepileptic, whereas its increase protects against seizure precipitation, is presumably too simplistic and certainly reflects only one side of the coin. It is now well known that GABAergic networks can promote and sustain synchronous cortical fast oscillations in the beta-gamma range.50 The observation that GABAA-receptor signaling can synchronize neuronal networks and can promote seizures may explain the disappointingly limited efficacy of antiepileptic drugs designed in the 1980s to potentiate GABA receptor–mediated actions. More subtle modulations of GABAA-receptor signaling should be envisioned to interpret the effects of currently available drugs and to develop new antiepileptic therapies. In line with this concept are the observation that bicarbonate efflux via GABAA-receptor channels61 is decreased by the carbonic anhydrase inhibitor topiramate and the demonstration of an antiseizure effect by the KCC2 antagonists that regulate chloride/potassium exchange. 62
MdC is supported by the Telethon Foundation grant GGP12265, by a grant of the Fondazione Banca del Monte di Lombardia, and by funds of the Italian Health Ministry (Finanziamento di Ricerca Corrente 2012–2015; Progetto Finalizzato Giovani Riceratori 2011-02348633). MA is supported by the Canadian Institutes of Health Research (CIHR grants 8109 and 74609) and by the Savoy Foundation. References for this Review were identified through searches of PubMed with terms “focal seizures,” “GABA,” “interneurons,” “epilepsy,” “extracellular potassium,” and “ictal and interictal discharges” up to March 2015. Further references were identified from those cited in articles. The final reference list was generated from articles that we considered relevant to the contents of our review.
Marco de Curtis, MD and neurologist, is director of the Epilepsy and Experimental Neurophysiology Unit at the Fondazione Istituto NeurologicoCarloBesta inMilano, Italy. He Chairs the ILAENeurobiology Commission for the period 2013–2017.
Massimo Avoli, MD, PhD, is a senior researcher at the Montreal Neurological Institute and Professor of Neurology, Neurosurgery, and Physiology at McGill University, Montreal, Quebec, Canada. He also teaches human physiology and neurophysiology at Sapienza University of Rome in Italy.
The authors declare no conflict of interest. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
Additional ContributionsMdC and MA equally contributed to the design and the writing of this review.