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/9780197549469.003.0057
Cognitive and behavioral impairment can drastically affect the quality of life of patients with epilepsy, particularly when seizures appear in early childhood. Treating such deficits is therefore critical, and understanding the mechanisms involved is an important step toward this goal. Although seizures are believed to play a critical role in these deficits, the underlying etiology of the epilepsy may also directly affect brain function. Unfortunately, dissociating the impact of seizure activity from the impact of etiology is complex. Experimental strategies developed to make this dissociation consist of three levels of investigation: behavioral performance analysis, neural coding properties of epileptic networks, and organization of neuronal dynamics through brain rhythms and oscillations. From this strategy, evidence suggests that both seizures and cognitive deficits take their origin in a discoordination of epileptic networks, caused by GABAergic dysfunction and/or the abnormal presence of excitatory loops. Such network discoordination manifests itself in the form of abnormal rhythmical activity or of epileptiform activity. Restoring such rhythmopathies might be the focus of future therapeutic perspectives.
In addition to seizures, patients with epilepsy also suffer from cognitive and behavioral impairment that can drastically affect the quality of life of patients and caregivers (Ronen et al., 2003; Loring et al., 2004; Berg et al., 2011). These deficits vary according to the epilepsy syndrome considered (Elger et al., 2004; Lin et al., 2012; Sokka et al., 2017) and, in a given syndrome, can also differ from one patient to another. These impairments, considered as comorbidities, comprise attention disorders, memory impairment, learning disability, hyperactivity, anxiety, depression, mental retardation, autism spectrum disorders, or social-adaptive deficits (Johnson et al., 2004; Epilepsies, Policy and Medicine, 2012; Lin et al., 2012).
Cognitive deficits in both children and adults with epilepsy are among the most concerning of the comorbidities (Hermann et al., 2002; Ronen et al., 2003; Berg, 2011; Holmes, 2015). Cognitive/behavioral deficits are greater when epilepsy begins in early childhood. Children with epilepsy have, on average, lower IQs and reduced school performance (Ellenberg et al., 1986; Aldenkamp et al., 1990; Fastenau et al., 2004, 2008), and a large proportion of them require special education services (Berg et al., 2011). Pharmacoresistant seizures before age 10 years are highly associated with long-standing cognitive impairment. In addition, epileptic encephalopathies (EEs), the most devastating epileptic syndromes, typically begin in early childhood (Berg et al., 2012). These observations indicate that seizures, or the etiologies causing the seizures, are particularly detrimental to developing neural circuits serving learning and memory processes.
Children with EEs are particularly worrisome in regard to cognitive comorbidities. The EEs are a group of disorders in which the unremitting epileptic activity contributes to severe cognitive and behavioral impairments above and beyond what might be expected from the underlying pathology alone, and these can worsen over time, leading to progressive cerebral dysfunction. The cognitive deficits in these children are often severe, with language, intellectual, and behavioral impairment (Nickels and Wirrell, 2017). Treating such comorbidities is therefore critical and understanding the mechanisms leading to them is an important step toward this goal.
Inherent in the concept of epileptic encephalopathy is the idea that suppression of epileptic activity may improve cognition and behavior. However, the etiology of the epilepsy clearly plays a significant role in the emergence of cognitive deficits (Epilepsies, Policy and Medicine, 2012; Korff et al., 2015; Sokka et al., 2017). Dissociating the impact of seizure activity from the impact of etiology is complex. In this chapter we will review clinical and experimental data that help make this dissociation. Since learning and memory deficits are a major cause of disability in patients with epilepsy, these disorders will be emphasized. Based on this data, we conclude that both seizures and cognitive deficits take their origin in a discoordination of epileptic networks, caused by GABAergic activity and/or the abnormal presence of excitatory loops. Such network discoordination manifests itself either in the form of abnormal rhythmical activity or of epileptiform activity. Restoring such rhythmopathies might be the focus of future therapeutic perspectives.
Of all the cognitive issues facing patients, memory impairment is one of the most common and disabling problems (Bell et al., 2011). There are two major kinds of memory, declarative (or explicit) memory and procedural memory, often referred to as implicit or nondeclarative memory (Zola-Morgan et al., 1983; Squire, 1986; Squire et al., 1990). Declarative memory is memory that can be demonstrated in some form and includes the conscious recollection of facts and events (Squire and Zola-Morgan, 1985, 1988). Spatial memory is the type of declarative memory responsible for encoding, storage, and retrieval of information regarding spatial orientation in the environment (Burgess et al., 2002). Spatial memory is the most commonly studied form of declarative memory in rodents (O’Keefe and Nadel, 1978; Buzsáki and Moser, 2013; Eichenbaum and Cohen, 2014).
Spatial coding can be egocentric (self-to-object) where the location of objects in space is relative to the body axes of the self and involves the dorsal striatum and connected structures. In allocentric (object-to-object) memory, the location of one object is defined relative to the location of other objects and is dependent on the hippocampus, entorhinal cortex (EC), and related structures (Scoville and Milner, 1957; Zola-Morgan et al., 1986; Banta Lavenex et al., 2014; Ekstrom et al., 2014; Hartley et al., 2014). Egocentric capabilities emerge prior to allocentric ones and dominate the infant’s spatial world for the first 6–12 months of life (Lavenex and Banta Lavenex, 2013). Allocentric spatial memory abilities in children emerge around 22 months of age (Rieser and Heiman, 1982; Bremner et al.,1994; Lew et al., 2000; Newcombe et al., 2007; Ribordy et al., 2013; Ribordy Lambert et al., 2017). Neuronal differentiation and synaptogenesis in the hippocampus reach adult levels at 3–5 years of age (Amaral and Dent, 1981; Ribak et al., 1985; Seress and Mrzljak, 1992; Seress and Ribak, 1995), corresponding to the time when allocentric memory appears (Baram et al., 2019).
To determine the extent of functional integrity in animal models, researchers have designed behavioral assays that test specific cognitive abilities and therefore distinct circuits. Rodents are excellent foragers and have remarkable navigational abilities. These abilities have been traditionally tested using a variety of assays (Fig. 57–1), including the Morris water-maze (Morris, 1981; Morris et al., 1982), Barnes maze (Barnes, 1979), radial arm maze (Olton and Samuelson, 1976), and active avoidance (Barry et al., 2016a,b ; see Mazarati et al., 2018, for a review centered on epilepsy). Testing spatial memory is particularly relevant for studying cognitive performance in temporal lobe epilepsy (TLE) and other models with memory deficits. However, these assays require the animal to find escape locations in aversive conditions such as being dropped in cold water or being exposed to open environments, which rodents tend to avoid. Such aversive components can be problematic in some epileptic models that are also affected by abnormal anxiety or fear levels. This can be, at least partially, avoided in other types of assays, relying on less stressful situations. Among these are the reaction-to-novelty tasks, which are based on the natural tendency of rodents to explore novel situations. By exposing the animal to an environment with different objects and then moving one of the objects to a new location, it is possible to induce an exploratory behavior oriented specifically to the new object location. Other assays can use positive reinforcements such as food rewards to bait specific locations in mazes of various shapes depending on the type of memory tested. For instance, working memory, relying on both hippocampus and prefrontal cortex, can be addressed by training animals to alternate between the arms of Y-, M-, or θ-shaped mazes for food rewards (Kim and Frank, 2009; Benchenane et al., 2010). Alternatively, working memory is also tested in delayed-match- (or non-match-) to-sample tasks taking place in operant conditioning boxes (Kleen et al., 2010, 2011).
Three levels of investigation to probe cognitive function in animal models. Behavior: In the Morris water-maze, rodents are place in a tank filled with opaque water. They have to find a platform hidden under the surfaces to escape. In the active avoidance (more...)
While tasks like the water-maze or Barnes maze help reveal spatial memory deficits, they may not sufficiently probe higher order memory abilities such as the ability to organize information in both space and time. This ability, also referred to as the “What-When-Where” memory is a fundamental component of episodic memory. To address this type of memory, more appropriate tasks such as the reaction-to-novelty tasks are particularly relevant. For example, in the context of epilepsy, the “What-When-Where” paradigm has been used to demonstrate specific episodic memory deficits in animals that had normal spatial performance in the water-maze (Inostroza et al., 2013). Unfortunately, higher level cognitive skills found in humans such as language and mathematics cannot be assessed in rodents.
One of the main issues with behavioral tests is that rodents can use various strategies to solve the same task (Packard and McGaugh, 1996). This can be problematic since cognitive deficits could be masked by the use of alternative strategies from the ones relying on the damaged brain structure. In animal models of epilepsy, where cell loss and structural alterations are commonly observed, this issue can be problematic. It can, however, be overcome by probing cognitive function directly through neuronal activity, in addition to behavior.
Neurons within the hippocampal formation are characterized by spatially modulated activity. In CA1, CA3, and the dentate gyrus (DG), principal (pyramidal) cells behave as “place cells” (Fig. 57–1); that is, they are active when the animal enters a specific region of the environment (the cell place field) and remain mostly silent in the remainder of the apparatus (O’Keefe and Dostrovsky, 1971). The place fields of different cells are distributed throughout the environment without specific relationship between the anatomical location of the cells within the hippocampus and the location of their place fields. While place fields remain stable across exposures to the same environment, the majority of place cells “remap” when the animal is placed in a different environment. Some place cells stop firing, new ones become active, and the remaining ones fire at different locations than in the previous environment. Interestingly, place cell stability relies on the same molecular and cellular substrates (NMDA receptors, protein synthesis) as the consolidation of episodic and spatial memory and long-term potentiation (LTP) (Kentros et al., 1998; Nakazawa et al., 2003; Agnihotri et al., 2004). Upstream of the hippocampus, medial entorhinal cortex (MEC) neurons, called grid cells, fire at regular intervals within the environment and form a triangular lattice (Fyhn et al., 2004). Other MEC neurons fire at the proximity of the walls and physical boundaries of the wall (Lever et al., 2009), according to the global heading direction of the animal, or a conjunction of position, direction, and velocity (Sargolini et al., 2006). Head direction cells can also be found in the posterior subiculum, the thalamus, and other subcortical nuclei (Taube et al., 1990). Evidence of place and grid cells have been found in humans and, as with rodents, encode for spatial location (Ekstrom et al., 2003; Doeller et al., 2010).
Altogether, the spatial reliability and stability of single-unit firing within the hippocampal formation make it a useful tool to investigate the underlying mechanisms responsible for cognitive deficits in epilepsy. By extension, neuronal coding properties of other sensory or integrative cortical regions could also provide useful information regarding information processing deficits in epilepsy.
Brain oscillations are rhythmical variations of the electroencephalogram (EEG) that reflect the repeated activation of synaptic currents in a structure (Anastassiou et al., 2016). The main oscillation patterns observed in humans and other mammals are δ (0.5–4 Hz), θ (4–8 Hz in human scalp EEG; 5–12 Hz in rodents), α (8–12 Hz in humans), β (15–30 Hz), γ (30–90 Hz), and high-frequency oscillations (HFOs: 90–300 Hz). Specific frequency bands are associated with particular behavioral states. In the hippocampus θ and γ are observed during exploration and rapid eye movement (REM) sleep (Vanderwolf, 1969; Bragin et al., 1995; Buzsaki, 2002). In contrast, automatic behaviors, such as grooming, immobility, or slow-wave sleep, are associated with HFOs, large irregular activity, or δ (Leung, 1980; Ylinen et al., 1995). Multiple brain rhythms can also interact with each other through phase and/or amplitude coupling. For instance, the amplitude of hippocampal γ increases during high θ periods (Bragin et al., 1995). γ also varies within θ cycles, reaching its maximum at the trough of θ (Fig. 57–1). This cross-frequency coupling is believed to provide higher order computational benefits (Buzsáki, 2010). Furthermore, neuronal activity can be modulated by two different oscillation frequencies, each being synchronized with different structure (Fujisawa and Buzsáki, 2011). Finally, the main brain rhythms and their relative frequency bands are preserved through evolution (Buzsáki et al., 2013). This not only confirms their functional relevance but also demonstrates that the neural substrates necessary for their genesis are common to multiple species.
There is accumulating evidence that brain rhythms play a critical role in information processing. First, by synchronizing the activity of input neurons with the membrane potential of target cells, oscillations provide optimal time windows during which neurons are more likely to be activated and therefore communicate. This communication through synchrony mechanism (Fries, 2005) binds the multiple structures that process diverse aspects about the same object, thereby enabling a cohesive experience. Second, by rhythmically increasing and decreasing neuronal excitability, oscillations segment the flow of information in packets or “chunks” that can be rapidly transmitted and interpreted to downstream structures (Gupta et al., 2012). These chunks (Fig. 57–2) consist of transiently recruited groups of neurons, called cell assemblies, that represent distinct cognitive entities. For instance, during θ cycles, which last for ~120 ms, place cells with overlapping fields are activated in compressed sequences that reproduce the ongoing trajectory of the rat, sweeping ahead of the animal (Gupta et al., 2012; Wikenheiser and Redish, 2015; Kay et al., 2020). From cycle to cycle, successive θ sequences progressively sweep further ahead of the animal, revealing planned trajectories. Third, characteristic alterations of brain rhythms and associated activities are observed in pathological situations, such as schizophrenia, depression, Alzheimer disease, and epilepsy (Uhlhaas and Singer, 2010; Buzsaki and Watson, 2012; Strüber and Herrmann, 2020; Lenck-Santini and Sakkaki, 2021).
Hippocampal sequences during exploration and rest. A. Schematic representation of the activity of different place cells (each with different colors) during exploration of a linear track (left) and subsequent rest (right). Top: The firing activity of (more...)
Three major oscillation patterns play a critical role in hippocampal function and therefore in episodic and spatial learning and memory: θ, γ, and HFOs.
In the hippocampus, θ oscillations reflect the slightly delayed, rhythmical activation of synaptic currents originating from EC and CA3 inputs in the stratum lacunosum moleculare (SLM) and stratum radiatum (SR), respectively. However, θ amplitude and frequency throughout the hippocampal formation are under the control of the medial septum/diagonal band of Broca (MSDB) (Buzsaki, 2002; Vertes et al., 2004), considered as the main θ pacemaker. Importantly, hippocampal neuronal firing is strongly modulated by θ activity, notably its phase, that is, where the oscillation is in regards to its peak or its trough. Strikingly, this influence is different according to the neuronal subclass considered: while pyramidal cells tend to fire action potentials at the trough of θ, Parvalbumin-expressing axo-axonic and basket cells fire action potentials preferentially at the peak and descending phase (Klausberger and Somogyi, 2008; Varga et al., 2012). In addition, place cells show a more precise phase precession of θ (O’Keefe and Recce, 1993; Skaggs et al., 1996). When the rat enters the place field of a cell, this cell will fire a burst of APs per θ cycle. As the animal crosses the place field, APs will appear at earlier θ phases for successive cycles, starting at the top of the ascending phase, falling to the trough, and exiting the field at the descending phase of the last θ cycle (Fig. 57–3A and B). At the population level, for a given θ cycle, place cells with overlapping place fields will therefore fire at different phases depending on the localization of the rat within their field. This ultimately results in the sweeping θ sequences mentioned above (Gupta et al., 2012; Wikenheiser and Redish, 2015; Kay et al., 2020). As with rodents, human hippocampal θ is elicited during navigation (Ekstrom et al., 2005; Bohbot et al., 2017). It is also observed while participants perform episodic and working memory tasks (Tesche and Karhu, 2000; Raghavachari et al., 2001, 2006; Rizzuto et al., 2003, 2006; Lega et al., 2012; Herweg et al., 2020) and during REM sleep (Cantero et al., 2003).
Phase precession. A. Place cell activity (vertical red bars represent its action potentials) superimposed with intrahippocampal LFP (black) filtered in the θ band (gray). When the rat crosses the place field of a cell (red disc), this cell will (more...)
γ oscillations in the hippocampus provide short (~30 ms) time windows within which cell assemblies form, that is, neurons that transiently bind together to process information (Harris et al., 2003). In CA1, at least two major, functionally distinct γ signals have been isolated. A slow (30–60 Hz) γ, located in the SR, is coupled in CA3 and a fast (60–90 Hz) γ in the SLM is coupled to EC (Colgin et al., 2009; Schomburg et al., 2014). These two γ signals can occur independently and therefore conveys different types of information (Fries, 2009; Bieri et al., 2014; Zheng et al., 2016). γ oscillations can also be generated locally within hippocampal circuits, notably through the strong influence of GABAergic interneurons (Whittington et al., 1995; Bartos et al., 2002; Cunningham et al., 2004; Bartos et al., 2007). In the hippocampus, γ is also strongly modulated by θ (Fig. 57–1): γ increases when θ is present and its amplitude varies as a function of θ phase (Bragin et al., 1995). Through the θ phase precession phenomenon explained above, the coordination between θ and γ appears to provide a coding scheme through which the representation of multiple items is sequentially ordered before being transmitted to connected structures (Lisman and Jensen, 2013; Fernández-Ruiz et al., 2017). In humans γ is recorded in most cortical areas, including the hippocampus and neocortex where it is believed to bind features encoded by several neuronal assemblies in multiple regions to form a coherent percept (Gray and Singer, 1989; Fries et al., 2007; Fries, 2015). As γ is altered in schizophrenia, this binding process is strongly impaired, therefore participating in its perceptual symptoms (Haenschel et al., 2009; Uhlhaas and Singer, 2010, 2015; Williams and Boksa, 2010).
HFOs (100–300 Hz), also referred to as “ripples” in the hippocampus (Stumpf, 1965; Buzsáki, 2015) are observed when rodents consume food, during immobility, drowsiness, or during slow-wave sleep (Buzsaki et al., 1992; Wilson and McNaughton, 1994). In Ca1, HFO amplitude is higher in the middle of the pyramidal cell layer and often co-occurs with large negative potentials in the SLM called “sharp-waves.” While the mechanisms responsible for sharp-wave/ripples (SPW-Rs) are still under investigation, it is likely that they are generated by a tonic excitation from CA3 coupled with the activation of a excitatory-inhibitory looped formed by pyramidal cells and perisomatic-targeting interneurons (Ylinen et al., 1995; Somogyi et al., 2014; Stark et al., 2014a). Evidence suggests that SPW-Rs play an important role in memory consolidation. For instance, during SPW-Rs, place cells are activated in a sequential order that is either the same or in reverse order (Fig. 57–2) of previous experiences (Foster and Wilson, 2006; Diba and Buzsáki, 2007; Gupta et al., 2010; Pfeiffer and Foster, 2013). In addition, exposure to high memory demand situations is associated with an increase of SPW-Rs, and artificially increasing their duration improves memory performance (Fernández-Ruiz et al., 2019). Interrupting awake SPW-Rs alters the long-term stabilization of place cells (Roux et al., 2017), and their disruption during sleep alters subsequent memory performance (Girardeau et al., 2009; Ego-Stengel and Wilson, 2010). Therefore, it is likely that place cell reactivation during SPW-Rs is involved in learning and memory processes. Coinciding with hippocampal ripples during slow-wave sleep (Siapas and Wilson, 1998), cortical sleep spindles are 7–14 Hz oscillations that last for 1–4 s that involve thalamocortical networks (Steriade et al., 1993; Contreras and Steriade, 1996; Steriade, 2000; Sirota and Buzsaki, 2005). Evidence suggests that spindles also play a critical role in learning and memory (Antony et al., 2018; Cairney et al., 2018).
In recordings from both patients and animal models, investigators have noticed the presence of abnormally fast HFOs, in the range of 250 to 600 Hz. Pathological HFOs (pHFOs) are considered an indicator of epileptogenicity (Bragin et al., 1999; Jacobs et al., 2012; Frauscher et al., 2017). The underlying mechanisms responsible for pHFOs are still under investigation (Jefferys et al., 2012). Among them is the emergence of network HFOs caused by the desynchronized, jittered firing of clusters of neurons (Ibarz et al., 2010).
In the previous section, various means to probe the integrity of cognitive function in both humans and animals models were presented. Though these cognitive makers, that is, behavior, neural coding, and brain rhythms, it is possible to investigate the output of the neural circuits, the major coding schemes, and neuronal dynamics. The diverse strategies used to investigate the cognitive effect of seizures, etiology, or both in animal models of epilepsy are presented.
In the past, considerable effort has been made to create animal models of epilepsy that mimic the seizure phenotypes and histological findings observed in patients. Until the recent advances in human genetics and the creation of transgenic mice, most models consisted of inducing a primary insult to the brain, either by kindling, chemoconvulsants, experimental head injury, or hypoxia/ischemia (Ben-Ari et al., 1979; Turski et al., 1983; Represa et al., 1989a; Pereira et al., 1992; Applegate et al., 1997; Leite et al., 2002; Löscher, 2011). To unravel critical mechanisms involved in epileptogenesis, these models were also associated with deficits in various cognitive modalities, including spatial, episodic, or working memory; social cognition; or executive function (Holmes et al., 1988; Dubé et al., 2006; Lenck-Santini and Holmes, 2008; Kleen et al., 2010, 2011; Talos et al., 2012; Inostroza et al., 2013; Barry et al., 2015). From these studies it is possible to conclude that the molecular, cellular, and structural changes induced by seizures may also result in cognitive deficits.
That said, there are several issues associated with extrapolating animal data to humans. The first issue concerns the damage caused by the methods used to create the models. It can be argued, for example, that these methods cause significantly more cell loss, inflammation, or plasticity abnormalities than those observed in patients. The second issue concerns the validity of the models used. Indeed, the presence of common phenotypes between patients and models is not enough to validate the model. There also needs to be similar molecular, cellular, and functional mechanisms. Finally, it is difficult to dissociate between the effects of seizures from those caused by the underlying etiology in these models. Various research groups have elegantly attempted to address these issues while revealing important physiological alterations responsible for cognitive deficits.
One of these strategies consists of inducing seizures while minimizing cell damage and restricting it to the hippocampal formation. For example, the kainic acid (KA) model used by Inostroza and colleagues (Inostroza et al., 2011, 2013) induces a milder cell loss than the pilocarpine, and the damage is restricted to the hippocampal formation. KA rats have normal anxiety levels and hypothalamus-pituitary-adrenocortical (HPA) axis function (Suárez et al., 2012), and their water-maze performance is comparable to controls (Inostroza et al., 2011). However, KA rats show specific deficits in episodic memory (Inostroza et al., 2013). Models that minimize cell damage are a valuable tool to investigate the mechanisms of epileptogenesis and comorbidities. Cell loss, together with neuronal reorganization (notably mossy fiber sprouting), is a hallmark of human TLE, and patients can have extensive hippocampal lesions. According to the extent of the lesion, one can expect to have varying degrees of impairment, including spatial and episodic memory deficits.
Another method used to induce seizures with resultant minimal cell loss is through the use of inhalant convulsive agents such as flurothyl, which induces generalized seizures only during the period of drug exposure (Applegate et al., 1997; H. Karnam et al., 2009; Niedecker et al., 2021). This method is particularly useful in controlling the number and frequency of seizures while avoiding mortality and long post-SE recovery. This method is particularly useful for rodent models of early-life seizures where pups can be returned to their mother minutes after the seizures. In immature rats, flurothyl-induced seizures do not cause cell loss but can lead to cognitive impairment, demonstrating that cell loss is not necessary for seizure-induced cognitive impairment (Karnam et al., 2009).
To test the impact of etiology without or with minimal seizures, it is possible to induce malformations of cortical development using the in-utero methyl-azoxy-methanol (MAM) injections. This model reproduces the neuropathology observed in patients with malformations of cortical development associated with early-onset epilepsy. MAM rats have neocortical and hippocampal malformations (Chevassus-Au-Louis et al., 1998; Rafiki et al., 1998), increased seizure susceptibility (Baraban and Schwartzkroin, 1996) but few spontaneous seizures (Harrington et al., 2007). These animals have important spatial learning and memory deficits. However, recurrent early-life seizures induced by flurothyl treatment had a limited impact on spatial performance (Lucas et al., 2011). These results suggest that the underlying malformation, not the seizures, is responsible for cognitive deficits in these animals. By extension, it can be argued that in patients with similar etiologies, aggressive antiseizure treatment may not be the most effective way to minimize cognitive deficits. As an alternative, environmental enrichment can provide significant benefits: it improves spatial performance (Jenks et al., 2013) and normalizes neuronal coding in MAM rats (Hernan et al., 2018).
More recently, advances in genetics and molecular biology allowed us to identify genes and cellular pathways involved in previously cryptogenic epilepsy syndromes. Either by completely knocking down the gene or introducing patient mutations in mice, it is now possible to reproduce epilepsy and/or cognitive phenotypes observed in these syndromes (Chen et al., 2004; Yu et al., 2006; Rossignol et al., 2013; Mistry et al., 2014; Milh et al., 2020). Interestingly, some of these models do not fully reproduce human seizure phenotypes but show cognitive deficits (Pederick et al., 2018). Improving these transgenic approaches, conditional methods, using Cre recombinase or the Tetracycline Response Element -Tetracycline transactivator (TRE-TTA) strategies allows investigators to induce genetic alterations in selected time periods and/or restrict genetic manipulations to specific cell types and/or time periods. Together with the use of viral vector injections, these strategies can be extremely valuable in understanding the mechanisms responsible for epilepsy and cognitive/behavioral phenotypes. Finally, RNA interference (RNAi), consisting of inserting, in neurons, double-stranded RNA sequences specific to the gene of interest induces a transient down-regulation of the downstream protein (Jacob, 2006). Contrarily to transgenic models that largely involve mice, this technique can be used in rats and does not require specific breeding. It also avoids cellular and structural compensation phenomena that occur in transgenic models, particularly during development. Finally, when restricted to a small portion of the brain, RNAi is not associated with confounding issues emerging when other parts of the brain and the body are affected. It is therefore particularly useful for investigating the functional role of a given gene/protein in a specific structure. These advantages are quite useful in dissociating etiology from seizures on cognitive function (Bender et al., 2013, 2016; Sakkaki et al. 2020).
To better understand the mechanisms responsible for memory impairment in TLE, Liu et al. (2003) recorded place cell activity in SE rats that had both spontaneous seizures and spatial memory deficits. Place fields from SE rats were less precise than controls and less stable across repeated exposures to the same environment (both at short and long term). In addition, they showed that a single seizure event, induced chemically, caused a transient cessation of place cell firing that was associated with a spatial performance deficit in the Morris water-maze. These results have recently confirmed by an elegant study from Shuman et al. (2020), who used, among other techniques, calcium imaging to investigate place cell activity in wireless, freely moving mice. Further confirming the impact of seizures on hippocampal function, H. Karnam et al. (2009) showed that recurrent, flurothy-induced seizures occurring during the first weeks of life caused long-standing alterations of place cell signal and stability. This effect was milder when seizures are administered in adulthood where they only cause a transient decrease of the number of place cells active in the environment while preserving their spatial quality (Lin et al., 2009). Altogether, these data suggest that the impact of seizures is age-dependent.
In rodent models of TLE, hippocampal θ amplitude and frequency are decreased (Dugladze et al., 2007; Chauvière et al., 2009; Richard et al., 2013; Inostroza et al., 2013; Shuman et al., 2017, 2020) even before the first spontaneous seizures. From the LFP, provided that multiple electrodes are used, it is possible to investigate the synchronization of oscillations between structures. Using this measure in the θ range, investigators have shown that the coordination between hippocampal subfields was abnormal in TLE rats (Laurent et al., 2015; Shuman et al., 2020). Notably, in KA-treated rats, (Laurent et al., 2015) showed that the temporal precision of θ oscillations between EC current sinks (from the temporo-amonic pathway and the perforant path) was decreased during exploratory theta. This deficit can be explained by the fact that, in their model, there is a selective cell loss in layer III of the MEC. Indeed, the decrease of θ coherence was restricted to the proximal part of CA1, where MEC projects, and counterbalanced by a coherence increase in the distal CA1, where LEC, intact in this model, projects. While such coherence deficits are likely specific to the model used, such a study shows that the precise temporal coordination of hippocampal circuitry is altered in epilepsy, even in models with restricted structural damage, such as the KA model. Selective episodic memory deficits are observed in these animals (Inostroza et al., 2013). At the single-unit level, we previously showed that phase precession is absent in a third of the place cells recorded in pilocarpine-treated rats (Lenck-Santini and Holmes, 2008). Interestingly, both normal and abnormal precession patterns are observed in simultaneously recorded cells, suggesting that precession deficits are caused by local network alterations rather than by an imbalance of global θ generators. Such local alterations likely originate from interneuron deficits. Indeed, it has been shown that the θ modulation of CA1 and dentate interneurons, but not principal cells, is altered in TLE models (Lopez-Pigozzi et al., 2016; Shuman et al., 2020). Interneurons show both a decrease in θ rhythmicity and a shift of the preferred theta phase at which they fire. That said, alterations of the EC (Kumar and Buckmaster, 2006; EA et al., 2007; Kumar et al., 2007; Laurent et al., 2015) and the MSDB (Garrido Sanabria et al., 2006), which directly project to hippocampal interneurons, may also participate in such discoordination. It is also worth noting that, intrinsic θ resonance properties of CA1 pyramidal cells are also altered in TLE models (Brewster et al., 2002; Jung et al., 2007; Marcelin et al., 2009; McClelland et al., 2011). Importantly, for cognitive comorbidities in TLE, θ alterations correlate to performance deficits in spatial, episodic, and working memory tasks (Chauvière et al., 2009; Richard et al., 2013; Inostroza et al., 2013; Barry et al., 2016). Finally, stimulation of the MSDB restores both theta oscillations and spatial performance in post-status epilepticu (SE) rats (Lee et al., 2017).
In addition to θ alterations, TLE is also associated with hippocampal γ deficits. In animal models, γ amplitude is decreased in the DG of pilocarpine-treated mice (Shuman et al., 2020) and in proximal CA1 of KA-treated rats (Lopez-Pigozzi et al., 2016). As noted earlier, there is a strong coupling of θ and γ oscillations within the hippocampus. Such coupling, which is likely under the control of GABAergic neurons, is strongly altered in TLE rats and correlates with episodic memory deficits (Lopez-Pigozzi et al., 2016).
As for other brain regions, the epileptic hippocampus can give rise to pHFOs. pHFOs can appear in conjunction with interictal spikes and SPW-R or independently (Bragin et al., 1999; Worrell et al., 2008; Ibarz et al., 2010; Jefferys et al., 2012; Alvarado-Rojas et al., 2015). Importantly, evidence suggests that they have a deleterious impact on cognitive function. Contrarily to normal HFOs, epileptic pHFOs can appear during locomotion, where they induce a decrease of theta amplitude and alter place cell coding (Ewell et al., 2019). They also can cause performance deficits when they occur during the inter-trial period of an episodic memory task (Kucewicz et al., 2014). Interestingly, by reducing pHFO frequency with carbamazepine and therefore restoring them to the normal HFO range, Valero et al. (2017) improved episodic memory performance.
Alterations of neural coding and brain rhythms are not specific to TLE. In children, the resting EEG is characterized by a greater low-frequency (delta to alpha) power. However, children with epilepsy tend to have even greater low-frequency activity (Michels et al., 2011). This excess in low-frequency power, sometimes coupled with decreased beta, is observed in idiopathic generalized epilepsy syndromes (Clemens et al., 2000) and involves different frequency bands and different cortical areas according to the type of syndrome considered (Clemens et al., 2012). While these changes may inform about the epileptogenicity of the underlying networks, evidence suggests that EEG alterations correlate with neuropsychological performance. For instance, Koop et al. (2005) showed that the presence of slow-wave activity in the EEG correlated with memory impairment in children with chronic epilepsy. Investigating 6- to 24-month-old infants with epilepsy, Kulandaivel and Holmes (2011) showed that, compared to children with normally developing infants, those with developmental delay had lower mean frequencies with higher delta and lower theta and alpha. This ratio was highly predictive of cognitive status. In patients with Dravet syndrome, alpha power starts to decrease between 3 and 5 years of age (Holmes et al., 2012). In a long-term follow-up study, Akiyama et al (2010) showed that the absence of background alpha in young adults correlated with worse intellectual disability. Auditory evoked γ responses are significantly reduced in patients with Dravet syndrome (Sanchez-Carpintero et al., 2020), suggesting that they may also have primary sensory deficits that may interfere with normal information processing.
In animal models, alterations of oscillations and neuronal coordination are also observed. For instance, in rats receiving early life seizures, changes in γ communication between the hippocampus and prefrontal cortices are observed and correlate with memory performance (Kleenet al., 2011; Mouchati et al., 2019; Niedecker et al., 2021). Rats that received early febrile SE also show alterations in CA3-CA1 γ and in Ca1 pyramidal cell theta modulation. Interestingly, these alterations correlate with the cognitive outcome of treated animals; that is, oscillation deficits are only observed in rats that perform poorly in a continuous spatial avoidance task (Barry et al., 2016b). Finally, transgenic mouse models of Dravet syndrome show altered cortical and hippocampal theta/γ coupling (Jansen et al., 2020), a reduction of hippocampal SPW-R frequency and intrinsic ripple frequency (Cheah et al., 2019), and a decrease in non-REM delta and REM theta power (Kalume et al., 2015).
Human premature EEG patterns are first characterized by transient bursts if rhythmic activity or sharp events intermixed with prolonged (from 5 s to minutes) periods of silence that decrease as pregnancy reaches its full term (Dreyfus-Brisac and Larroche, 1971; Khazipov and Luhmann, 2006). These bursts of activity are dominated by brief (<6 s) slow 0.3–2 Hz activity that appears alone or superimposed with faster, 8–25 Hz spindle-like rhythm (delta brushes). While some discontinuity is still present at full term, EEG patterns are fully elaborated at this age and electrographic sleep cycles are present (Dreyfus-Brisac and Monod, 1975; Scher, MS, 2008). Strikingly, an activity similar to the human premature patterns (discontinuous activity and delta-brushes) is present in neonatal rodents in vivo (Khazipov et al., 2004; Khazipov and Luhmann, 2006; Milh et al., 2007). In the neocortex of newborn rats and mice, intracranial recordings show transient 5–25 Hz activity that lasts for approximately 1 second and that are called spindle bursts. Associated with neocortical spindle bursts are hippocampal early sharp waves, consisting of large negative potentials, highest in the SLM, that can be followed by 0.5–1.5 s bursts of beta (10–30 Hz) oscillations. Early sharp waves and spindle bursts are, in a large majority, preceded by myoclonic twitches. By postnatal day (P) 9–12, these patterns disappear and are replaced by oscillation patterns similar to those seen in adults.
By 2 years of age, the EEG resembles that of an adolescent, with a well-developed, reactive alpha rhythm of 7–8 Hz, distinct centro-parietal rhythms, and well-formed frontally dominant fast activity. Sleep spindles, which play an important functional role in sleep-dependent synaptic plasticity and memory consolidation (Fogel and Smith, 2011), are well developed by 4 years (McClain et al., 2016).
The ontogeny of spatial memory in rodents is highly orchestrated with critical periods where axons and dendrites establish appropriate connections that optimize information processing across broad networks (Sheperd, Sheperd, 2018AD; Shepherd, GM, 2018; Sporns et al., 2004). As mentioned above, the medial EC creates a neural representation of space through functionally distinguished cell types: grid cells, border cells, head direction cells, and speed cells (Sargolini et al., 2006; van Strien et al., 2009; Rowland et al., 2016) with the rate of maturation unique to each type of neuron (Langston et al., 2010; Ainge and Langston, 2012; Wills et al., 2014; Tan et al., 2017). Grid cells first emerge around P21 and develop functional properties rapidly. They possess most of the properties that characterize adult grid cell firing at the time they are first detected (Langston et al., 2010; Wills et al., 2010, 2012). In vitro recordings show that mEC grid cell network synchronization significantly increases at P22 (Langston et al., 2010). This suggests that the widespread recurrent excitatory network necessary for grid cell activity emerges at this age (Fuhs and Touretzky, 2006; McNaughton et al., 2006; Burak and Fiete, 2009). Place cells, which can be identified earlier than grid cells mature gradually but at a slower pace than grid cells (Langston et al., 2010; Wills et al., 2010; Scott et al., 2011). It is proposed that the spatially tuned firing of grid and place cells arises from local computations as a result of the specific synaptic architecture of the network (McNaughton et al., 2006; Couey et al., 2013; Moser et al., 2014). Patterned activity propagating through the mEC-hippocampal circuit during this sensitive developmental phase finely tunes synaptic connectivity of the network and drives the emergence of specific firing patterns (Kropff and Treves, 2008; Muessig et al., 2015). At each stage of the circuit, excitatory activity drives the maturation of downstream areas of the network in a linear and directional developmental sequence (Donato et al., 2017). Interestingly, during SPW-R, place cells replay depicts stationary locations until P19–P21, when entire trajectories start to be represented. Compression of temporal sequences during theta cycles only appear at P23, when grid cells are mature. Altogether, these data suggest that, during the critical period for memory development, MEC-hippocampal activity provides the driving force behind the temporal coordination on neuronal ensembles underpinning spatial memory (Farooq and Dragoi, 2019; Middleton and McHugh, 2019; Muessig et al., 2019).
Whereas normal activity patterns are required for circuit maturation, aberrant neuronal activity can disrupt spatial cognition. This is particularly important in the case of massive bursts of synchronized network activity which occur during seizures. Recurrent (Liu et al., 1999; H. B. Karnam et al., 2009; H. Karnam et al., 2009; Holmes et al., 2015) or prolonged seizures (Dube et al., 2009; Barry et al., 2015; Barry et al., 2016, 2020; Patterson et al., 2017) during the critical period also have long-standing effects on spatial cognition, hippocampal rhythms, and place cell firing patterns. However, disruption of hippocampal rhythms by other means can also result in cognitive impairment. To evaluate the effects of disruption of normal hippocampal rhythms during the critical period on allocentric memory in rodents, Kloc et al. (2020) used optogenetic stimulation of the medial septum to introduce randomly varying hippocampal oscillations from 1 to 110 Hz for 1 to 5 hours a day during the critical period of spatial memory development (P21–P25). Rats receiving disruptive hippocampal stimulation during the critical period for memory development for either 1 h or 5 h had marked impairment in spatial learning as measured in an active avoidance test of spatial cognition compared to control rats. Thus, seizures during the critical period may affect spatial cognition by disrupting the normal ontogeny of hippocampal rhythms.
Importantly, direct or indirect alterations of GABAergic activity are associated with a significant proportion of epilepsy syndromes with severe cognitive impairment. In TLE, in addition to mossy-fiber sprouting, there is a significant modification of hippocampal GABAergic function caused by a loss of specific classes of interneurons or synaptic reorganization of the surviving ones (de Lanerolle et al., 1989; R Cossart et al., 2001; R. Cossart et al., 2001; Andrioli et al., 2007; Wyeth et al., 2010; Drexel et al., 2012; Marx et al., 2013; Tóth and Maglóczky, 2014). In upstream structures, like the EC and medial septum, GABAergic neurons and their innervation are also affected in TLE models (Garrido Sanabria et al., 2006; Kumar et al., 2007). In other types of epilepsies, including epilepsies of genetic origin, alterations of GABAergic function are common. This is the case in mutations of GABAergic receptors (Jiang et al., 2016; Oyrer et al., 2018) but also when the excitability of GABAergic neurons, particularly the fast spiking ones is preferentially decreased, as observed in mutations affecting sodium (Nav1.1) or calcium (Cav2.1) channels (Yu et al., 2006; Rossignol et al., 2013). Finally, some malformations of cortical development affect the proliferation, migration, or intrinsic properties of inteneurons (see Represa, 2019, for a review). In view of the large number of GABAergic alterations in epilepsy, the term “interneuronopathies” referring to those syndromes “that are associated with impaired development, migration, or function of interneurons” has been adopted (Kato and Dobyns, 2005; Katsarou et al., 2017). Importantly, most of these syndromes are associated with both epilepsy and neurodevelopmental disorders, including epileptic encephalopathies where the cognitive state of the patients is severely altered (Brooks-Kayal, 2011).
In addition to their role in the epileptogenic process, GABAergic neurons are also critical for the temporal coordination of neural networks and oscillations which themselves play a fundamental role in cognitive functions. GABAergic neurons within the hippocampus and MSDB are directly involved in the generation and regulation of theta rhythms (Goutagny et al., 2008; Hangya et al., 2009) and gamma rhythms (Whittington et al., 1995; Cunningham et al., 2004; Marlene Bartos et al., 2007; Cardin, 2018); the interaction between GABAergic and glutamatergic cells is critically involved in the coordination of hippocampal ripples (Stark et al., 2014b). Finally, LFP patterns coordinating the thalamo-cortical networks during sleep are also supported by GABAergic neurons (Paz and Huguenard, 2015). Therefore, alterations of GABAergic function in epilepsy may also play a direct role in cognitive deficits. While the concept of excitation-inhibition balance is a fundamental principle in ictogenesis, it can also be extended to a more dynamic view of brain function where precisely coordinated interaction between excitatory and inhibitory neurons is essential for information processing. This dual role of GABAergic function in epilepsy and cognitive function may explain the apparent correlation between seizure frequency/severity and cognitive deficits in some syndromes. Under this perspective, both phenotypes may have the same origin: a deficit of dynamic systems or, as others have coined, a “rhythmopathy” (Lopez-Pigozzi et al., 2016; Shuman et al., 2017).
The functional role of GABAergic activity in adult networks is likely to be carried out in developing circuits. There are, however, specificities about the immature brain that may render it even more important. Indeed, GABAergic activity during perinatal and postnatal development plays a critical role in the establishment of neural circuits. During this period, GABAergic cells control the survival, migration, and synaptic and axonal connectivity of both GABAergic and glutamatergic neurons (LeMagueresse & Monyer al., 2013; Duan et al., 2020; Bortone and Polleux, 2009; De Marco García et al., 2015; Tuncdemir et al., 2016; Butt et al., 2017). This control acts through synaptic signaling but also via a paracrine action (Demarque et al., 2002; Manent et al., 2005). Furthermore, the neonatal period is dominated by spontaneous network activity. GABAergic cells that are born the earliest play a critical role in orchestrating the activity of the network (Bonifazi et al., 2009a/b; Picardo et al., 2011; Bocchio et al., 2020). Sensory deprivation during critical periods induces GABAergic cell death and strongly alters the functional properties of cortical circuits (Fox, 1992; Simons and Land, 1994; Micheva and Beaulieu, 1995; Modol et al., 2020). Altogether these data suggest that abnormal GABAergic function during perinatal development has dramatic and permanent effects on the future functional properties of cortical networks. Coupled to seizures and their downstream physiological alterations, these effects may be potentiated.
The rhythmopathy hypothesis discussed above implies that cognitive deficits and seizures have a common pathophysiological mechanism. Therefore, it may be difficult to dissect out the adverse effects of seizures on cognition from the cognitive deficits induced by the rhythmopathy.
Dravet syndrome, an epilepsy syndrome of infancy characterized by severe, recurrent, and pharmacoresistant seizures, learning disability, autistic features, and profound cognitive deficits (Catterall et al., 2010; Dravet, 2011; Han et al., 2012), provides an example of how dissociation between seizures and rhythmopathy may be addressed. The EEG of children with Dravet syndrome is characterized by an age-dependent slowing of α oscillations toward the θ band (Holmes et al., 2012). In adults, the absence of occipital background α rhythm correlates with higher intellectual disability scores (Akiyama et al., 2010). In contrast, there is no statistical relationship between the severity/frequency of seizures and cognitive deficits (Nabbout et al., 2013). This suggests that oscillation alterations in Dravet syndrome are significant contributors to cognitive dysfunction.
Dravet syndrome is caused, in a majority of cases (~80%), by pathogenic variants in the SCN1a gene coding for the alpha1 subunit of the voltage-gated sodium channel (Nav1.1) (Catterall et al., 2010). The mechanism of Dravet syndrome linked to Scn1a is haploinsufficiency (Scheffer and Nabbout, 2019). Most mutations are de novo but can be inherited in ~10% of the cases, from mosaic parents (Myers et al., 2018). In 2015, more than 1200 pathogenic variants of the Scn1a gene have been detected (Meng et al., 2015). While mutations causing a loss of function of the protein cause Dravet syndrome, other variants are also responsible for milder epilepsy syndromes such as generalized epilepsy with febrile seizures (Escayg et al., 2000; Meisler et al., 2021).
Nav1.1 is expressed at higher proportions in fast-spiking GABAergic neurons, and its alterations appear to selectively impair the firing properties of this class of cells (Yu et al., 2006; Tai et al., 2014). Conditional knockout restricted to GABAergic cells is sufficient to induce severe epilepsies, and this phenotype is even stronger than when the KO also involves pyramidal cells (Dutton et al., 2013). Therefore, Dravet syndrome could be considered as an interneuronopathy. Interestingly, Nav1.1 alterations are also implicated with nonepilepsy syndromes, such as autism spectrum disorders and Alzheimer disease (Verret et al., 2012; D’Gama et al., 2015; Sawyer et al., 2016; Martinez-Losa et al., 2018). Martinez-Losa et al. (2018) showed that overexpression of Nav1.1 selectively in interneurons restored brain rhythms and cognitive performance in hAPP transgenic mice. Finally, recent experiments in Scn1a mice show that hippocampal theta-gamma coupling is altered and that SPW-R frequency and intrinsic ripple frequency are reduced (Cheah et al., 2019; Jansen et al., 2020). In total, these results suggest that GABAergic coordination of networks and oscillations is also altered in Dravet syndrome and may be directly involved in cognitive deficits. However, to test this hypothesis, one needs to have a model that does not present seizures and avoids confounding factors associated with Nav1.1 deficits, such as ataxia or risks of SUDEP, compensatory mechanisms, or potential alterations of brain development. We therefore developed an ShRNA approach to down-regulate, via RNA interference, the expression of NAv1.1.
By injecting, in adults, ShRNA vectors in specific anatomical targets, we could avoid these confounders. In a series of experiments we showed that (Bender et al., 2013, 2016) Nav1.1. downregulation in the MSDB disrupted fast-firing activities in this structure and decreased the downstream θ in the hippocampus. This manipulation caused performance deficits in spatial working memory and reference memory tasks that correlated with θ alterations. Importantly, animals that underwent the MSDB Scn1a ShRNA injections did not have seizures.
To determine how local hippocampal networks are affected by Nav1.1.downregulation, we later injected Scn1a ShRNA vectors in a restricted (1mm2 area), unilateral portion of the dorsal CA1 (Sakkaki et al., 2020). Nav1.1 downregulation caused a decrease of the firing rate of putative interneurons, but not pyramidal cells, and a reduced θ/γ coupling and shift of the preferred θ phase at which pyramidal cells fire. The spatial accuracy of Scn1a SHRNA place cells was lower than controls, and their θ phase precession and compression of ongoing sequences were abnormal (Fig. 57–3C). Finally, Scn1a SHRNA-injected rats also had spatial memory deficits, despite the restricted extent of the injections. Therefore, independently of seizures or other confounding factors, Nav1.1 alterations are responsible for deficits in the coordination of both global and local networks. Evidently, seizures in Dravet syndrome are not to be discarded since they also can lead to more cognitive deficits, alterations of the network, and increase the risk of SUDEP (Massey et al., 2014; Dutton et al., 2017; Salgueiro-Pereira et al., 2019).
In this review, we presented strategies to investigate mechanisms responsible for cognitive deficits in epilepsy. Understanding these mechanisms is complex, notably because of the difficulty of distinguishing between the impact of the etiology of the syndromes and the consequences of epileptic activity on network function. To address this issue, we highlighted the relevance of three levels of investigation: analysis of behavior through tests designed to probe specific functions; neural coding properties, which represent the implementation of these functions; and neuronal dynamics or brain rhythms, which coordinate information processing and communication within and across structures. Dissociating the impact of etiology and seizures on cognition has also been possible through the use of different experimental strategies in various models, ranging from seizure induction in wild-type animals to reproducing syndrome etiologies while avoiding seizures. While epileptiform activity, particularly in the developing brain, likely contributes to cognitive deficits, we propose that alterations of the temporal coordination of brain networks, caused in a majority of syndromes by alterations of GABAergic function, are also significant contributors to these deficits. In addition, it is suggested that seizures and functional deficits are both consequences of the breakdown of temporal dynamics in epilepsy syndromes. Therapeutic strategies, focused on restoring the dynamic coordination of epileptic networks, may improve both seizure and cognitive outcomes.
Important questions remain to be answered. First, how can temporal coordination be restored? When GABAergic function is altered, this could be achieved through grafts of embryonic GABAergic progenitors, which functionally integrate to the network and successfully restore seizures and cognitive function (Baraban et al., 2009; Hunt et al., 2013; Hsieh and Baraban, 2017). The next challenge is to understand the functional mechanisms behind such improvement; specifically which cell type, connectivity, and rhythmical properties are most relevant. GABAergic function could also be cell specific, when deficit of a particular cell type is identified (Lupien-Meilleur et al., 2021). Of course, developing treatment strategies using implants or optogenetic or chemogenetic approaches in humans will be challenging.
Epilepsy is not uniquely defined by GABAergic dysfunction, and excessive excitability can also participate in pathogenic processes. In TLE, for instance, the presence of mossy fiber sprouting, that is, abnormal, recurrent, excitatory loops in DG-CA3 networks, has been identified as a major contributor to epileptogenesis and probably cognitive deficits (Scheibel et al., 1974; Tauck and Nadler, 1985; Represa et al., 1989b; Crepel et al., 1997; Buckmaster, 2010; Artinian et al., 2011). In this process, granule cells make new, abnormal excitatory connections to other dentate granule cells, CA2 and CA3 pyramidal cells (Scharfman et al., 2003) that recruit kainate receptors not normally present (Epsztein et al., 2005). These synapses have slower kinetics than traditional granule cell synapses (Crépel and Mulle, 2015) and disorganize dentate granule cell dynamics (Artinian et al., 2011), hereby causing seizures (Peret et al., 2014) and likely alter DG information processing. Restoring normal DG kinetics, through modulation of KA receptors, may therefore improve both seizure and cognitive outcome.
The majority of epilepsy syndromes associated with profound cognitive/behavioral impairment concern those starting in early childhood. However, most of what we know on the mechanisms responsible for cognitive deficits comes from TLE and early-life seizures. Future efforts should therefore concentrate on models of developmental epilepsies and particularly on EEs. In EEs, it is critical to further determine whether and how alterations of GABAergic function during critical periods of development affect the establishment and maturation of cortical networks. Attention should be paid to the heterogeneity of GABAergic neurons and the developmental period at which these neurons contribute to the fate of cortical networks (Bonifazi et al., 2009a/b; Picardo et al., 2011; Rossignol et al., 2013; Jiang et al., 2018). As a consequence, treatment strategies focused on specific classes of interneurons and tailored to their specific developmental windows may be critical (Lupien-Meilleur et al., 2021). Understanding the involvement of GABAergic neurons in pathogenic processes is made difficult by the fact that seizures, also caused by GABAergic dysfunction, affect early networks. The use of conditional knockout, knock-in, and RNAi, coupled to new techniques like optogenetics and chemogenetics, may help make this distinction.
This work was supported by a grant from the National Institutes of Health R01NS076763 to PPLS, NS108765 and NS108296 to G.L.H.
The authors declare 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...
