Childhood absence epilepsy (CAE) is one of the most common genetic forms of epilepsy, with a typical presentation of behavioral arrest associated with 3–4 Hz generalized spike-wave discharges (SWDs), occurring frequently throughout the day (Crunelli and Leresche, 2002). Animal models of CAE and absence seizures have evolved significantly over the last 70 years, giving incremental insight into the developmental pathophysiology of CAE with each additional model (Avoli, 2012). The first phase consisted of penicillin-induced spike-wave seizures in the feline generalized epilepsy model where very basic aspects of thalamocortical oscillations were uncovered using convulsant drugs (Avoli, 1995; Kostopoulos, 2017). The second phase was driven by the important discovery of spontaneous genetic models and their single-cell neurophysiology in rodents (Maheshwari and Noebels, 2014; Fig. 21–1).

Figure 21–1.

Timeline of milestone studies investigating the pathophysiology of absence epilepsy. ASM = anti-seizure medication; C = cortex; CC = corpus callosum; CN = cerebellar nucleus; EEG = electroencephalogram; EPSP = excitatory postsynaptic potential; FF = (more...)

We are now in the third stage, with the powerful capability of engineering conditional mutations in specific genes throughout the network and the ability to monitor and manipulate large-scale, cell-type-specific ensembles in awake and unanesthetized behaving mice. These recent breakthroughs have produced a significant shift in the understanding of the corticothalamic circuit dysfunction in absence epilepsy and promise to help us identify key molecular and cellular targets for therapy. Coupled with the recent ability to identify individuals with epilepsy due to mutations in these exact genes, this next-generation approach should accelerate our understanding of absence epilepsy as well as our ability to better diagnose and treat the disorder. Given the emerging genetic heterogeneity of this disorder, our goal in this chapter is to review the early mechanistic discoveries and then turn to the most recent studies that define common downstream mechanisms that may be shared by the nearly 40 monogenic forms. In particular, impaired cortical and thalamic synaptic inhibition, most prominently mediated by parvalbumin-expressing, fast-spiking interneurons (PV-INs), has become increasingly implicated in the pathogenesis of absence epilepsy over time.

Brief History of Thalamocortical Inhibition in Absence Epilepsy

In 1949, building on earlier descriptions of spontaneous and evoked “petit mal” attacks recorded in the electroencephalogram (EEG) of humans and animals (Gibbs et al., 1935), Hunter and Jasper reported that 10–30 Hz electrical stimulation of the intralaminar nuclei in the thalamus, but not sensory relay nuclei, in unanesthetized cats could produce an “arrest reaction” with 3 Hz SWDs in cortical and thalamic recordings, whereas 200 Hz stimulation produced SWDs followed by generalized convulsive seizures (Hunter and Jasper, 1949). Since glutamatergic intralaminar nuclei project widely to diffuse cortical regions, this observation supported the “centrencephalic pacemaker” hypothesis, in which the thalamus was the generator for the excitatory cortical and thalamic discharges observed in absence epilepsy. Li and Jasper reported later the first extracellular single-unit recordings in cat cortex during SWDs induced by local application of strychnine (Li and Jasper, 1953), a glycine receptor antagonist (Breitinger and Breitinger, 2020). Notably, they immobilized unanesthetized animals with D-tubocurarine, which can act as a convulsant in cortex, highlighting the need to take into account possible confounding factors with studies performed under immobilization. The additional confounder of barbiturate anesthesia, which enhances GABAergic inhibition (Henschel et al., 2008; Löscher and Rogawski, 2012), also does not align well with results obtained in awake animals. Regardless, this was the first evidence of cellular firing patterns coinciding with the spike component of the SWD on the EEG, in a pharmacological animal model of cortical disinhibition.

Further studies continued to show cortical synaptic disinhibition as a common mechanism to induce SWDs. Marcus and Watson showed that application of either strychnine or 10% pentylenetetrazol (PTZ), a GABA_A receptor antagonist, to the cortex alone was sufficient to elicit generalized 3 Hz SWDs, and that disconnection of thalamic afferents did not prevent these phenomena (Marcus, 1966). Similarly, Prince and Farrell showed that parenteral administration of penicillin (another GABA_A receptor antagonist) in cats produced widespread 4.5 Hz SWDs, most prominently visible in association cortex (Prince and Farrell, 1969). Additional studies (Fisher and Prince, 1977; Kostopoulos and Avoli, 1983) in the feline penicillin-induced generalized epilepsy (FGPE) model demonstrated that cortical EPSP-IPSP sequences recorded from both pyramidal tract and nonpyramidal tract cortical neurons mirrored spike and wave phases of the absence seizure EEG patterns. The importance of cortical circuitry was recognized as transcortical section disrupted cortical synchrony during SWDs more than corticothalamic lesions. However, since reduced thalamic activity suppressed generalized penicillin-induced feline SWDs (Avoli and Gloor, 1981), and likewise cortical inhibition suppressed SWDs on the EEG in the same feline model (Avoli and Gloor, 1982), both cortical and thalamic hyperactivity were felt to be necessary to generate and sustain the rhythmic discharges characteristic of spike-wave complexes. A gradual increase in cortical responsiveness to thalamocortical stimulation, particularly from diffuse thalamic projections, was identified as an underlying mediator of penicillin-induced SWDs (Kostopoulos and Avoli, 1983). Again, perturbation of both cortical and thalamic inhibition was found to influence the pathogenesis of absence seizures.

Importantly, there is a distinction between the GABA_A-mediated disinhibition, which leads to the initiation of absence seizures, and the pathophysiology behind the maintenance of intra-seizure rebound bursting. Giaretta, Avoli, and Gloor used intracellular deep-layer cortical recordings in the FGPE model to show that somatic postsynaptic inhibitory activity in motor cortex (Giaretta et al., 1987) was not reduced during absence seizures. Previously this had been assumed to be a major factor during both the initiation and maintenance of seizures, as seen in focal seizures after local penicillin injection, albeit under barbiturate anesthesia (van Duijn et al., 1973).

Further work elucidated that the EEG wave component of SWDs was not generated by GABA-mediated IPSPs since neuronal input resistance is strongly increased, but rather by a combination of the calcium-activated potassium current (I_K(Ca)) and, to a lesser extent, sodium-activated potassium current (I_K(Na)) following each depolarized spike; and the spike component was mainly facilitated by hyperpolarization-activated, cyclic-nucleotide gated (HCN) channel currents (I_h) (Charpier et al., 1999; Timofeev et al., 2000, 2004; Timofeev and Steriade, 2004). Based on previous work by Llinas and Jahnsen (Jahnsen and Llinás, 1984; Llinás and Jahnsen, 1982) and on the observation that thalamocortical relay neurons can generate rhythmic oscillations at hyperpolarized membrane potential levels (McCormick and Prince, 1988), McCormick and Pape used thalamic slice recordings from cats (as well as guinea pigs) to characterize the I_h current mediated by HCN channels in thalamic relay neurons. They concluded that these channels, in conjunction with the low-threshold T-type Ca²⁺ channels, can cause 1–2 Hz high-frequency burst patterns associated with reduced attention and sleep (McCormick and Pape, 1990), laying the groundwork for the involvement of these same channels in absence seizures.

Toward the end of the last millennium, a series of studies by Mircea Steriade and collaborators concluded that (1) local bicuculline injection in the cortex but not the thalamus produced SWDs, but thalamocortical activation can hypersynchronize cortical activity if intracortical synaptic connections are inactive (Steriade and Contreras, 1998); (2) intracellular deep-layer cortical recordings revealed fast-rhythmic-bursting (FRB) cells which were highly synchronized to EEG spikes, suggesting a role in synchronization and initiation of SWDs (Steriade et al., 1998); (3) simultaneous dual recordings of neurons showed strong cortical synchronization during bicuculline-induced SWDs in anesthetized cats, and dual cortical and thalamic local field potential (LFP) recordings revealed that the thalamus lagged the cortex at seizure onset (Neckelmann et al., 1998); and (4) many thalamocortical cells were hyperpolarized during SWDs, but some displayed rebound bursting at the same frequency as cortical neurons during SWDs, similar to thalamocortical slice studies showing selective amplification of 3 Hz stimuli (Timofeev et al., 1998). The mystery that remained was how disinhibition due to GABA_A receptor inhibition resulted in the cyclic recruitment of HCN and T-type calcium channels in both cortex and thalamus. Further work in rodent models of absence epilepsy shed light on this quandary by demonstrating the important role of increased tonic inhibition (Fig. 21–3).

Figure 21–3.

Pathophysiology of absence epilepsy. Monogenic mutations leading to absence epilepsy fit into a general scheme with entry points in decreasing fast feed-forward inhibition, increasing tonic inhibition and de-inactivation of T-type calcium channels. (more...)

Evolution of Available Models of Absence Epilepsy

The thalamic components of SWD oscillatory loop activity became the focus of several lines of investigation in the 1980s and 1990s, after the centrencephalic origin of absence seizures had been postulated in the 1950s. Notably, two novel genetically undefined inbred rat models approximating human absence dynamics, albeit with SWDs emerging in adulthood and at 8–14 Hz oscillation frequency versus 3 Hz typically seen in humans (Marescaux et al., 1984; van Luijtelaar and Coenen, 1986; Vergnes et al., 1982), were used to hypothesize that PV-INs in the reticular thalamic nuclei (RTN) may be capable of inducing thalamocortical oscillations and SWDs via a combination of the I_h (McCormick and Pape, 1990), Na⁺, and low-threshold T-type Ca²⁺ currents (Avanzini et al., 1993, 1989; Coulter et al., 1990, 1989; Tsakiridou et al., 1995). Specifically, pharmacological blockade of the Ca²⁺ current in the RTN was shown to be sufficient for significant reduction of SWDs in GAERS rats (Avanzini et al., 1993). In addition, T-type Ca²⁺ currents in RTN were shown to grow abnormally large during development as seizures emerge in Genetic Absence Epilepsy Rats from Strasbourg (GAERS) (Tsakiridou et al. 1995). Finally, this low-threshold Ca²⁺ current in the ventrobasal nucleus of the thalamus was found to be blocked by clinically therapeutic anti-absence drugs, including ethosuximide and dimethadione (Coulter et al., 1990, 1989). Together, these experiments solidified the role of abnormally activated T-type calcium channels as the final common pathway of expression of absence seizures, which made blockade of seizure activity by T-type calcium channels with ethosuximide a key criterion for future models of absence epilepsy.

Discovery of the tottering mouse, the first single locus mutant rodent with absence epilepsy (Noebels and Sidman, 1979), was followed by related spontaneous monogenic mutant models of the 1990s, including three mouse lines characterized around this time, the stargazer mouse (Noebels et al., 1990), the lethargic mouse (Burgess et al., 1997), and the ducky mouse (Barclay et al., 2001). These models provided stronger phenotypical resemblance to the human disease with an earlier onset of SWDs, as well as an opportunity to precisely map the effects of single gene mutations on neurophysiological and behavioral phenotypes, specifically the Cacng2 gene encoding an AMPA receptor trafficking protein (stargazer), the Cacnb4 gene encoding a subunit of P/Q-type calcium channels (lethargic), and the Cacna1a gene encoding the α_1A subunit gene encoding voltage-sensitive P/Q-type calcium channels (tottering, Fletcher et al., 1996). However, the relationship between these genetic mutations and the pathogenesis of absence epilepsy was not yet well understood at the time of their discovery.

By the late 2010s, thanks to genetic engineering, the list of monogenic rodent models of absence seizures had grown substantially to 16 calcium channel mutations, 4 glutamate receptor-related channel mutations, 5 GABA receptor channel mutations, and 7 other monogenic models (Maheshwari and Noebels, 2014). Other more recent SWD models include the TRIP8b knockout mouse model (Heuermann et al., 2016), the ANK3 knockout mouse model (Lopez et al., 2017), and the neuroligin-2 knockout mouse model (Cao et al., 2020). These genetic models lend strong support to the underlying hypothesis that dysfunction in the feedforward inhibitory circuit (either corticothalamic involving the RTN or thalamocortical involving cortical PV-INs) has the potential to lead to absence epilepsy (Maheshwari and Noebels, 2014).

Debate on the Thalamic versus Cortical Onset of Seizures

To try and pinpoint initiation sites of absence seizures in rat models more precisely, several studies in the 2000s examined activity spread with more sophisticated recording technologies. Using high-density EEG and simultaneous thalamic LFP recordings, perioral sensory cortex was found to consistently lead other cortical areas and the thalamus during the first 0.5 sec of spontaneously occurring SWDs (Meeren et al., 2002).

Further evidence has been collected in support of a cortical focus for the initiation of absence seizures. In 2004, it was found that focal application of ethosuximide directly to peri-oral S1 cortex was just as effective at suppressing SWDs as systemic administration (Manning et al., 2004). In 2007, Polack used in vivo whole-cell recordings to show that deep-layer somatosensory cortical (SSC) neurons consistently lead more superficial neurons at the initiation of SWDs, are generally hyperactive compared to controls, and spontaneously produce 10 Hz single-cell (individual) oscillations interictally and preictally (Polack et al., 2007). The same year, a developmental study of WAG/Rij rats showed that these animals significantly lose HCN1 channel expression, and in turn I_h current strength, as SWDs develop. Dual somato-dendritic patch-clamp recordings in S1 cortex showed that this leads to increased intrinsic layer 5 cortical bursting activity, consistent with the cortical focus hypothesis (Kole et al., 2007).

In another publication by Polack, evidence for SSC as the SWD initiation site in GAERS was further solidified since inactivation of SSC by local TTX injection effectively abolished SWDs (Polack et al., 2009). Furthermore, recurrent activation of GABAergic interneurons in the cortical focus was found to control the bursting behavior of ictogenic neurons (Chipaux et al., 2011). In addition, GAERS SSC was found to have progressively increasing excitability, synaptic activity, and synchronized oscillations as seizures develop (Jarre et al., 2017). Finally, when expression of P/Q type calcium channels in mouse cortical layer VI projection neurons was selectively suppressed, stereotypical spontaneous 5–7 Hz SWDs appeared that were sensitive to ethosuximide and coincided with behavioral arrest (Bomben et al., 2016). This was mediated by elevated T-type calcium currents in postsynaptic thalamic relay and reticular cells, demonstrating that this single mutation in cortical neurons can chronically remodel thalamic excitability leading to generalized SWDs. While these studies emphasize the pivotal role of primary SSC as the natural initiation site of SWDs in the GAERS model, it is important to note that this is not the only location in the corticothalamic circuit that can be vulnerable to switching the circuit’s behavior from the normal state to the SWD state. For example, electrical stimulation of secondary SSC and insular cortex can trigger SWDs, and in this model, insular cortex was found to lead S1 (Zheng et al., 2012).

Weir and Sie pointed out in 1966 that there is clear evidence for multiple possible initiation sites for SWDs that generate the same EEG discharge patterns (Weir and Sie, 1966). Owing to the differences in the species, behavioral states, and recording and manipulation techniques used over the decades, it is not surprising that several theories on the origin of the stereotypical SWD pattern have been formulated. While there has been widespread consensus that the primary and essential components of this circuitry are found in the interplay between cortical and thalamic networks, attempts to prove that a single site is necessary to initiate SWDs in a particular model has been challenged by evidence that SWDs can be initiated in alternative sites, even in the same model. Therefore, while there are many potential initiation sites between the thalamus and the cortex depending on the animal model and recording condition, the initiation of any given SWD may be dependent on the site with the lowest relative threshold at that moment. Accordingly, studies of human CAE have recently revealed dynamic epileptogenic networks with one or several hyperexcitable cortical nodes responsible for SWD initiation and maintenance in response to physiological thalamic afferent activity, with thalamic structures amplifying and propagating the oscillation (Kokkinos et al., 2017).

Several recent review articles have provided comprehensive summaries of the literature on human and animal studies of absence epilepsy. Among others, Depaulis and Charpier emphasize the prominent role of SSC as an entry point for the generation of aberrant corticothalamic oscillations (Depaulis and Charpier, 2018). Huguenard highlights in his recent review that both thalamus and cortex are sufficient to induce SWDs, and that these pathological oscillations are an embedded feature of the network that is effectively suppressed in healthy individuals (Huguenard, 2020).

In order to integrate evidence from more recent studies of SWD initiation mechanisms, the existence of a cortical initiation network (CIN), consisting of many possible sites of abnormal preictal cortical activity, has been proposed. The significance of the CIN rests in the conclusion that cortico-thalamo-cortical networks are inherently predisposed to the emergence of SWD patterns which can be triggered by a multitude of entry points (Crunelli et al., 2020). However, one class of entry points that appears to be shared by more experimental models than others involves a disturbance in corticothalamic feedforward inhibition (FFI) (Maheshwari and Noebels, 2014; McCafferty et al., 2018; Panthi and Leitch, 2019). Furthermore, the importance of corticothalamic FFI was reinforced by biophysical modeling that demonstrated its role in regulating thalamic oscillation frequencies contributing to SWD generation (Chen et al., 2017; Fan et al., 2017).

Next, we focus on evidence highlighting the cell-type-specific role of FFI in the generation of SWDs in rodent models (Fig. 21–2). Much of the insight previously gained into the pathophysiology of absence epilepsy has come from experiments using genetic and molecular techniques. Strategies to generate absence seizures in rodents include monogenic mutants, inbred rats, and drug provocation (e.g., bicuculline, gamma-hydroxybutyrate). Each of these strategies has shed light on cell-type-specific contributions to absence epilepsy (Table 21–1).

Figure 21–2.

A generic model of disynaptic fast feedforward inhibition (FFI). This model applies to both thalamocortical and corticothalamic FFI. Long-range monosynaptic excitation of excitatory neurons (E) in the thalamus and cortex are moderated by disynaptic inhibition (more...)

Table 21–1

Insights from Selected Models of Absence Epilepsy.

Insights from Drug-Induced Models of Absence Seizures

Drug-induced models of absence seizures have the advantage of comparing the wild-type state to the absence seizure state in the same animal but lack construct validity since the drug-induced seizures are only temporary. The drugs that can induce absence seizures in rodents can be broadly separated into drugs that reduce phasic inhibition (synaptic GABA_A antagonists) and drugs that increase tonic inhibition (directly via extrasynaptic GABA_A agonists or indirectly via GAT-1 antagonists or synaptic GABA_B agonists). GABA_A antagonists that induce absence seizures include bicuculline (Matejovská et al., 1998), penicillin (Ostojić et al., 1997), PTZ (McLean et al., 2004; Snead, 1992a); and GABA_B agonists that induce absence seizures include gamma-hydroxybutyrate (GHB) and its prodrug gamma-butyrolactone (Snead, 1988). In addition, GABA_B agonists such as baclofen exacerbate GABA_A antagonist-induced absence seizures (Snead, 1992b). Extrasynaptic GABA_A receptors are the main source of tonic inhibition and can also be activated by the extrasynaptic GABA_A receptor agonist, 4,5,6,7-tetrahydroisoxazolo[5,4-c]pyridin-3-ol (THIP), to induce absence seizures (Depaulis, 1992). GABA_B receptor agonists indirectly increase tonic inhibition by recruiting more extrasynaptic GABA_A receptors (Cope et al., 2009). Finally, tonic inhibition can also be increased indirectly by inhibiting GAT-1, the GABA transporter which clears GABA from the extrasynaptic space. Altogether, drug-induced models of absence seizures reveal that either reduction in fast phasic inhibition or enhancement of slow, tonic inhibition can reproduce the electroclinical features of absence seizures. These features form the basic framework underlying absence epileptogenesis (Fig. 21–3) and are reinforced by data derived from monogenic models.

Insights from Monogenic Models

Monogenic mouse models of absence epilepsy provide several unique aspects which allow greater insight into the pathogenesis of absence epilepsy. First, the onset of spontaneous seizures is most often in the second postnatal week, approximating the onset of absence seizures in children with CAE. Second, the product of the involved gene can shed light onto the specific pathways that may be affected in its absence. Finally, the presence of wild-type littermate controls allows direct testing of hypotheses against a control animal that has a similar genetic background but should not have absence epilepsy.

Several lines of evidence from the genetic mouse models point to dysfunction specific to neocortical PV-INs and PV-INs of the thalamus (RTN). In the corticothalamic feedforward inhibitory circuit, there is a selective loss of activation of RTN neurons in Gria4 mutant mice (Paz et al., 2011). Similarly, there is reduced AMPA receptor subunit expression in RTN in stargazer mice (Barad et al., 2012). Both of these studies show that reduced phasic activation of RTN PV-INs can lead to absence epilepsy. On the neocortical side, there is specific expression of stargazin in neocortical PV-INs in wild-type mice, which is absent in stargazer mutant mice, associated with reduced AMPA-receptor trafficking in these neurons (Maheshwari et al., 2013). These findings were corroborated using immunogold staining in conjunction with electron microscopy showing reduced AMPA receptor subunit expression in cortical PV-INs in stargazer mice (Adotevi and Leitch, 2017). In another genetically distinct model of absence epilepsy, Cav2.2 channels could compensate for loss of Cav2.1 (Cacna1a) in somatostatin-expressing interneurons (SST-INs), but GABA release was significantly impaired when Cav2.1 was removed from PV-INs (Rossignol et al., 2013). These findings indicate that genetic mutations leading to absence epilepsy can cause dysfunction of PV-INs in both neocortex and RTN, resulting in loss of fast FFI in both thalamocortical and corticothalamic circuits, respectively. Monogenic models have also contributed to understanding the role of increased tonic inhibition and enhanced T-type Ca²⁺ channel current (Maheshwari and Noebels, 2014).

Further Insight from Inbred Rat Models of Absence Epilepsy

In the two most common inbred rat models (WAG/Rij, GAERS), there is further evidence supporting dysfunction in feedforward inhibitory circuits. In GAERS, there is reduced GABA_A expression in both sensorimotor cortex and anterior thalamic regions (Spreafico et al., 1993). In WAG/Rij rats, there is reduced peak conductance of fast neocortical IPSPs (D’Antuono et al., 2006) as well as reduced PV-immunoreactivity in somatosensory cortex (van Luijtelaar and Sitnikova, 2006). Altogether, these findings support an underlying loss of cortical and thalamic FFI. Based on data from the monogenic models of absence epilepsy, the findings in these inbred rat models are more likely to be primary pathological processes, rather than compensatory phenomena. However, further work is necessary to fully disentangle primary and secondary changes.

A Working Model of Absence Epileptogenesis

Altogether, findings from drug-induced models, monogenic models, and inbred rat models of absence epilepsy converge on a series of interconnected processes that can lead to the generation of spontaneous absence seizures (Fig. 21–3). Excitatory communication between the cortical and thalamic nodes of the corticothalamic loop is normally regulated by fast FFI such that an EPSP in the target excitatory neuron is followed by an IPSP within a few milliseconds. The disynaptic inhibitory circuit, centered around a fast-spiking, parvalbumin-expressing interneuron, can be interrupted anywhere from the presynaptic release of glutamate from the excitatory neuron of one node to the postsynaptic transmission of GABA onto the excitatory neuron of the other node (Maheshwari and Noebels, 2014). This loss of FFI can be exogenously mimicked by GABA_A receptor antagonists. Disinhibition leads to hyperexcitable neurons that recruit feedback inhibition from local inhibitory neurons which increases tonic inhibition of excitatory neurons (Cope et al., 2005). Dysfunction of astrocytic GABA uptake via GAT-1, for reasons that remain unknown, compound the degree of tonic inhibition (Cope et al., 2009). Increased tonic inhibition of fast-spiking neocortical neurons can also theoretically further exacerbate SWDs by causing further reduction in FFI (Krook-Magnuson et al., 2008). The gradual increase in tonic inhibition leads to de-inactivation of T-type calcium channels, which then promote SWDs with cyclic burst firing of both excitatory and inhibitory neurons. The degree to which cortical and thalamic excitatory and inhibitory neurons are engaged in these SWDs in vivo has only recently been elucidated with more advanced in vivo imaging and cell-type-specific manipulation of neuronal activity.

In Vivo Imaging Studies

Recent advances in microscope technology have enabled neuroscientists to record fluorescent optical signals corresponding to electrical activity from large neuronal populations in vivo, across many different species. These techniques, including in vivo two-photon calcium imaging, widefield imaging of voltage-sensitive dyes, and other modalities, have found widespread acceptance in a growing array of epilepsy-related lines of research (Rossi et al., 2018; Somarowthu and Goldberg, 2020; Wykes et al., 2019). While many of these recent experiments have centered on models of focal or generalized convulsive seizures (Somarowthu et al., 2021), the following section highlights recent publications using in vivo imaging studying animal models of absence epilepsy.

In a pivotal study using PTZ to induce absence seizures, voltage-sensitive dye imaging (VSDI) showed moderately increased spontaneous cortical activity but not thalamic activity in γ2R34Q mutants (a GABA_A receptor subunit mutant model of absence epilepsy) during interictal periods (Witsch et al., 2015). VSDI data furthermore showed that whisker-evoked activity was able to spread over larger cortical areas in supragranular primary SSC after PTZ injection. These experiments support both a cortical initiation network and a general deficit of cortical inhibitory drive. However, the interpretation of these findings is somewhat limited since mice were both anesthetized with urethane and required PTZ to generate the absence seizures.

These limitations were overcome in a study of the stargazer mouse model of absence epilepsy, where spontaneous seizures were recorded with EEG while imaging visual cortex in awake animals using two-photon calcium imaging. Activity was generally suppressed during seizures and pair-wise correlation coefficients were also reduced (Fig. 21–4). Interestingly, cortical neuronal activity on average dropped several seconds before seizure onset, but recovered with a much shorter lag after seizure termination (Meyer et al., 2018). Furthermore, a subset of PV-INs and SST-INs were found to also be overall suppressed during absence seizures. These results indicated that (1) large cortical neuronal populations are not uniformly engaged during rhythmic ictal activity, even though 6–9 Hz rhythmic discharges can be measured ubiquitously using surface EEG electrodes; (2) preictal reduction in unit activity before EEG discharge onset indicates early engagement of these neurons during SWDs; and (3) the high variability in the engagement of neurons indicates that SWDs are routinely generated by diverse ensembles of neurons from seizure to seizure.

Figure 21–4.

Two-photon imaging of absence epilepsy. a. Stargazer visual cortex, layer 2/3, ensemble of neurons labeled via intracortical adult AAV injection with the calcium indicator GCaMP6m. Scale bar = 50 µm. Red arrow = neuron with hyperactivity during (more...)

In contrast to an optical imaging method, a recent study of seizure-associated cortical activity patterns used high-density EEG arrays (36 channels plus 2 thalamus electrodes) to record and visualize SWD activity. They recorded from two different mouse models (GHB and PLCb4) and found seizures to be variably initiated in the cortex and never originated in thalamus in both models (Lee et al., 2019). This study demonstrated the utility of employing diverse methodologies for investigations of absence epilepsy—in this case using a method with relatively low spatial and high temporal resolution, as opposed to the higher spatial but lower temporal resolution achieved with two-photon imaging.

Introducing another variation of nonpenetrating electrical recording modalities, impedance tomography has been used to reconstruct activity spread during SWDs in the cortex at a maximal signal depth of 2 mm (Hannan et al., 2018). Upon electrically evoked SWD initiation in anesthetized Sprague-Dawley rats, a sharp drop in tissue impedance was consistently found to propagate in the dorsoventral direction. These findings confirmed that the barrel cortex was a low-threshold site for SWD initiation.

It is worth noting that recent studies have shown an impressive diversity of approaches to investigate and visualize abnormal brain activity patterns associated with absence epilepsy. At the same time, many mechanistic questions about the etiology of absence epilepsy remain to be solved. While each individual type of in vivo imaging modality remains limited in some respects, these methods offer unique advantages over other approaches and are certain to continue generating informative findings, such as further deciphering the role of specific neuronal cell types and pathways whose discoveries may aid in the development of novel precision medical treatments.

Optogenetic/DREADD Studies

Relative to in vivo imaging methods, recently developed optogenetic and chemogenetic methods of neuronal activity manipulation have been adopted in epilepsy research at an even faster pace. There have been a number of excellent reviews summarizing the application of these techniques across all types of epilepsy (Cela and Sjöström, 2019; Christenson Wick and Krook-Magnuson, 2018; Lieb et al., 2019; Walker and Kullmann, 2020). Here, we focus on studies that specifically evaluated interventions to ameliorate absence epilepsy-related seizure activity.

In WAG/Rij rats, 10 Hz optogenetic stimulation of cortical pyramidal cells in S1 caused SWDs in 10% of trials. This group used an optrode multi-electrode array (MEA) design introduced in an early proof-of-concept study demonstrating the utility of simultaneous optical stimulation and single-unit multichannel electrical recordings (Wang et al., 2010). The seizure induction probability depended on LFP power before stimulation and response amplitude during stimulation. Evoked waves of activity stimulated a cortical focus which then sustained SWDs beyond the stimulation duration (Wagner et al., 2014). While this study showed that optogenetic stimulation of cortical pyramidal cells can induce SWDs, more specific perturbation within the corticothalamic circuit was investigated in later studies.

In contrast, optogenetic stimulation of subcortical pathways aborts SWDs, but it has proven broadly ineffective in triggering them. In another study using optogenetics, tottering and C3H/HeOuJ mice were used to record single cerebellar nucleus (CN) cells and EEG in awake animals. AAV2-hSyn-ChR2(H134R)-EYFP was introduced for optogenetic stimulation of CN neurons, both unilaterally and bilaterally with 30–300 msec light pulses, activating neurons consistently. Notably, while stimulation did not trigger seizures, single light pulses were sufficient to abort SWDs within 500 msec (Kros et al., 2015). The proposed mechanism of aborting seizures is through excitation of hyperpolarized thalamic neurons, preventing activation of HCN channels and the deinactivation of T-type calcium channels. Given the effectiveness of cerebellar outflow stimulation in these two models with different underlying mechanisms, this potential new approach for intervention appears worth exploring further.

In another study, Sprague-Dawley rats were intraperitoneally injected with Gamma-butyrolactone to generate SWDs. Here, rAAV5-hSyn-ChR2(H134R)-mCherry was injected in the deep/intermediate layers of the superior colliculus. SWDs were reduced with unmodulated light stimulation, but more effectively with 5 Hz pulsed light (Soper et al., 2016). The authors speculate that this effect could be mediated via activation of the pedunculopontine nucleus, which projects widely to both cortex and thalamus. Remote activation of corticothalamic neurons may therefore prevent hyperpolarization-mediated burst firing within the absence seizure loop.

Multiple optogenetic approaches, including inhibitory variants and sustained-acting opsins, have also been used to study and interrupt SWD oscillations (Sorokin et al., 2017). Using halorhodopsin to suppress excitatory ventrobasal thalamocortical neurons results in postinhibitory rebound bursting, SWDs were generated in both stargazer mice and WAG-Rij rats. To achieve the opposite effect, stabilized step function opsins (SSFOs) were then used to depolarize TC neurons and switch them from phasic to tonic firing. This desynchronized cortical oscillations and reduced SWDs. In a related experiment, optogenetic stimulation of PV-INs in the RTN was shown to be sufficient to terminate SWDs, whereas stimulating SST-INs had no effect (Clemente-Perez et al., 2017). Therefore, whereas hyperpolarization of excitatory thalamocortical relay neurons may generally induce SWDs, this effect would not likely be caused by elevated inhibitory input from PV-INs of the RTN.

Another thalamocortical circuit element that appears to be important in regulating rhythmic activity is neuroligin 2 (NLG2), expressed in ventrobasal thalamic neurons (VB) but not in the RTN or cortex. Neuroligin traffics GABA_A receptors to the synapse (Cao et al., 2020). In NLG2 knockout mice, SWDs are believed to be initiated via reduced feedforward inhibitory postsynaptic activity in the VB leading to thalamocortical hyperactivation. However, extrasynaptic GABA_B receptors are not impaired in NLG2 knockout mice and are active in the maintenance phase of SWDs by contributing to burst firing from increased tonic inhibition. To show that enhancing GABAergic transmission at the RTN-VN synapse can suppress SWDs, they introduced ChR2 under the Dlx promoter (expressed in both PV-INs and SST-INs) bilaterally in RTN and activated their terminals in VB thalamus with blue light in 5-sec intervals. During light stimulation, the proportion of time spent in SWDs was half that under control conditions; thus, in both of these optogenetic rescue experiments, reversal of the feedforward inhibitory dysfunction was associated with activation of PV-INs to some degree.

In comparison with optogenetic activation or suppression, “designer receptors exclusively activated by designer drugs” (DREADDs) have certain advantages and shortcomings. Activation or inhibition over long periods of time can be maintained until the activating ligand is no longer given to the animal, without the need for more invasive application of light. The effectiveness does not depend on light intensity, and the size of the impacted population only depends on the expression of the receptors, not a combination of the optogenetic construct and the penetration depth of the light. On the other hand, temporal control is limited, and light intensity is easier to titrate than the effective ligand concentration in the brain. One can also combine multiple optogenetic actuators sensitive for different wavelengths to activate or suppress the same populations. To date, only two studies utilized DREADDs for the investigation of absence epilepsy. In the first study, the authors crossed a stop-floxxed inhibitory DREADD mouse line with animals expressing Cre in PV-INs to suppress activity in these cells. The required DREADD ligand, CNO, was either given systemically or via local injections in SSC or thalamus. This consistently generated SWDs and behavioral arrest, demonstrating the crucial role of this specific population of PV-INs (Panthi and Leitch, 2019). Subsequently, work from the same laboratory showed that activation of FFI mediating PV-INs can effectively suppress absence seizures induced by systemic injection of pentylenetetrazol (PTZ). PV-Cre expressing mice were crossed with hM3Dq-flox mice, such that excitatory DREADDs were activated upon local injection in either SSC or reticular thalamic nuclei. In both cohorts, DREADD activation proved effective and thus exemplified a potential new therapeutic approach of targeting specific cell populations for treatment of absence epilepsy (Panthi and Leitch, 2021).

Altogether, these studies show the significant potential of optogenetic and chemogenetic manipulation to both dissect pathological network mechanisms, and test novel approaches for targeted, cell-type specific interventions. It is worth noting that together with more biological information, more theoretical and computational modeling work will be needed in order to explain the diversity of experimental results from different genetic absence models (Ge et al., 2019). Some recent studies have begun to apply sophisticated computational methods to re-create thalamocortical network dynamics and their transitions from physiological to pathological patterns in silico (Deeba et al., 2018; Fan et al., 2017; Yang and Robinson, 2017). This way, new hypotheses can be generated, results can be generalized and translated between models, and theories unified that might bring the field closer to novel clinical approaches for more effective treatment of absence epilepsy and its comorbidities than what is currently available.

We would like to thank Jeffrey Noebels, MD, PhD, for his mentorship throughout both of our careers.

We have no disclosures.