Can rodent models elucidate the pathomechanisms of genetic epilepsy?

Autosomal dominant sleep‐related hypermotor epilepsy (ADSHE; previously autosomal dominant nocturnal frontal lobe epilepsy, ADNFLE), originally reported in 1994, was the first distinct genetic epilepsy shown to be caused by CHNRA4 mutation. In the past two decades, we have identified several functional abnormalities of mutant ion channels and their associated transmissions using several experiments involving single‐cell and genetic animal (rodent) models. Currently, epileptologists understand that functional abnormalities underlying epileptogenesis/ictogenesis in humans and rodents are more complicated than previously believed and that the function of mutant molecules alone cannot contribute to the development of epileptogenesis/ictogenesis but play important roles in the development of epileptogenesis/ictogenesis through formation of abnormalities in various other transmission systems before epilepsy onset. Based on our recent findings using genetic rat ADSHE models, harbouring Chrna4 mutant, corresponding to human S284L‐mutant CRHNA4, this review proposes a hypothesis associated with tripartite synaptic transmission in ADSHE pathomechanisms induced by mutant ACh receptors. LINKED ARTICLES This article is part of a themed issue on Building Bridges in Neuropharmacology. To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v179.8/issuetoc

Over the past two decades, such experiments using single-cell and KI mice models of ADSHE have provided some elucidation of ADSHE pathogenesis, though even more contradictions have subsequently emerged. However, exploring the functional abnormalities of the typical ADSHE-mutant α4 and β2 subunits of nAChR has remained a significant target in understanding ADSHE pathomechanisms (Becchetti et al., 2015;Boillot & Baulac, 2016).
Recently, a part of the outline of ADSHE pathomechanisms was clarified using a transgenic rat model bearing S286L-mutant rat Chrna4, corresponding to the S284L-mutation in human CRHNA4 of patients with ADSHE. However, the identified functional abnormalities associated with ADSHE were also more complicated than expected. Complex pathomechanism cascades of ADSHE originate from functional abnormalities of mutant α4β2-nAChR, from alterations of its cation channel to wide-range abnormalities, involving unexpected intracellular signalling (K. Fukuyama, Fukuzawa, Okubo et al., 2020;Fukuyama et al., 2020b;Fukuyama, Ueda et al., 2020). These experimental findings, which reveal various functional abnormalities originating from mutant α4β2-nAChR, also elucidated the pathogenesis of the age-dependent onset and clinical features of the ADSHE subclass, including comorbid cognitive impairment and anticonvulsant sensitivity (K. Fukuyama, Fukuzawa, Okubo et al., 2020;Fukuyama et al., 2020b;Fukuyama, Ueda et al., 2020;Picard et al., 1999;Yamamura et al., 2013;Zhu et al., 2008). Full elucidation of the complex pathophysiological cascade associated with ADSHE will contribute to uncovering the mechanisms of genetic epilepsies caused by other gene mutations. Furthermore, it may lead to breakthroughs in understanding the common pathomechanisms of focal epilepsies. Therefore, this review discusses the pathomechanisms of ADSHE, focusing on S284L-mutant ADSHE, comparing the functional abnormalities using single-cell and KI mouse models bearing other classical ADSHE mutations, such as S280F, insL and V287L.

| CLINICAL FEATURES OF ADSHE/SHE SYNDROME
The clinical manifestations of ADSHE and sporadic SHE are indistinguishable, because they are comparable to frontal lobe epilepsy and usually occur during non-rapid eye movement (non-REM) sleep (Provini et al., 1999;Scheffer et al., 1995;Tinuper et al., 2016). SHE is a rare form of focal epilepsy, with an estimated prevalence of 1.8/100,000, thus fulfilling the definition of a rare disease . ADSHE seizures are sleep-related, stereotyped hypermotor seizures consisting of vigorous hyperkinetic features or asymmetric dystonic/tonic features. In addition, many paroxysmal arousals are observed. More rarely in some patients, seizures can also manifest as epileptic nocturnal wandering. Patients usually experience a cluster of hypermotor seizures during the same night. Electroencephalogram (EEG) often fails to detect ictal discharge during hypermotor seizures . These brief seizures may sometimes evolve into secondary generalised tonic-clonic seizures .
Even if ADSHE seizures are controlled, once a patient experiences a seizure, they can experience clustering and many/frequent ADSHE seizures during the same night (Provini et al., 1999;Scheffer et al., 1995). Based on these clinical findings, it has been deemed that ADSHE seizures (sleep-related complex, stereotyped hypermotor seizures) probably represent the spectrum of SHE syndrome, despite differences in EEG sensitivities (Montagna, 1992;Provini et al., 1999;Tinuper et al., 2016).
Traditionally, ADNFLE seizures were classified into three typical types: nocturnal paroxysmal arousals (NPA), nocturnal paroxysmal dystonia (NPD) and episodic nocturnal wandering (ENW) (Provini et al., 1999). Clinically, the terminologies describing NPA, NPD and ENW are currently outdated in patients with ADSHE. However, these three typical ADNFLE seizures are the major clinical terminologies, at the generation of genetic ADNFLE rodent models. Therefore, in this review, the epileptic phenotypes of genetic ADNFLE rodent models were defined by NPA, NPD and ENW. ADSHE/SHE are classified into two subclasses based on anticonvulsant sensitivity and cognitive dysfunction (Picard et al., 1999;Tinuper et al., 2016). Genetic variations appear to be closely associated with the clinical features of subclass formation (Table 1).
From a different perspective, another achievement of single-cell models was that some ADSHE mutations, such as S280F-and insLmutant α4β2-nAChRs, displayed enhanced carbamazepine sensitivity (Bertrand et al., 2002;Picard et al., 1999), whereas the S284L-mutant α4β2-nAChR exhibited reduced carbamazepine sensitivity (Bertrand et al., 2002). The correlation between carbamazepine sensitivity with the clinical and electrophysiological features of ADSHE-mutant α4β2-nAChRs suggests that these mutations are possibly involved in the pathophysiology of ADSHE subclasses, even if it is unclear if the abnormality of ADSHE-mutant α4β2-nAChR results in a loss of function or gain of function (Bertrand et al., 2002).

| FUNCTIONAL ABNORMALITIES IN RODENT MODELS
The transfected single-cell model with the patch-clamp method is a useful technique to elucidate the functional abnormalities of ion channels. However, neuroscientific experiments using epileptic animal models can identify the responsible neural circuits and clarify the various developmental processes of epileptogenesis/ictogenesis. Epileptic animal models are practical tools to explore the pathogenesis/pathophysiology of age-dependent onset and can trigger event-related features of genetic epilepsies. Nonetheless, the human brain is  Yamamura et al., 2013;Zhu et al., 2008). Unfortunately, there have been no consensus validation criteria for genetic epilepsy models, except for the spontaneous absence epilepsy models, GAERS and WAG/Rij (Coenen & Van Luijtelaar, 2003;Yamamura et al., 2013). This review only introduces our validation criteria, which are partial modifications of the validation criteria of Coenen and van Luijtelaar (Coenen & Van Luijtelaar, 2003;Fukuyama et al., 2020a;Okada et al., 2010;Yamamura et al., 2013;Zhu et al., 2008).
• Face validity of an ADSHE model is the ability to faithfully mimic the clinical symptoms of the disorder, including NPA, NPD, ENW and clustering during non-REM sleep (Fukuyama et al., 2020a;Okada et al., 2010;Zhu et al., 2008).
• Predictive validity is the ability to predict previously unknown mechanisms of medication sensitivity of ADSHE, showing anticonvulsant sensitivity comparable to those of ADSHE patients (Fukuyama et al., 2020a;Okada et al., 2010;Zhu et al., 2008).
• Construct validity of ADSHE conforms to a theoretical rationale for genetic epilepsy, such as gene mutation, expression of mutant molecules or transmission abnormalities (Fukuyama et al., 2020a;Okada et al., 2010;Zhu et al., 2008). Coenen and Van Luijtelaar (2003) listed the electroencephalographic features, typical spike-and-wave discharges, in the construct validity of absence epilepsy models, because the epileptic discharges in cortico-thalamo-cortical network play important roles in the generation of simultaneous onsets of phenotypic (absence seizure) and electroencephalographic (spike-and-wave discharge) phenomena in both patients and rodent models with absence epilepsy (Coenen & Van Luijtelaar, 2003 Predictive validity (or pharmacological validity) seems to be a highly vexed concept. Traditionally, one of the major targets of approved anticonvulsants, voltage-dependent Na + channels (VDSC), has been identified post hoc by studying the mechanism of action of drugs identified by serendipity. Additionally, absence seizures can be elicited by a relatively low dose of pentylenetetrazole in the absence epilepsy models, compared to healthy rodents, and this model has been used in the preclinical screening of putative antiepileptic drugs (Chen et al., 2011;Coenen & Van Luijtelaar, 2003). Therefore, phar- contributing to predictive validity but should be considered as contributing to construct validity. This is because the identified functional abnormality and its response to treatment might become a candidate target for development of novel medications.

| S280F-and insL-mutant models
Two KI animal models harbouring the same mutant Chrna4 gene, corresponding to the human S280F-mutant CRHNA4, have been generated and are known as S252F-KI (Klaassen et al., 2006) and S248F-KI (Teper et al., 2007) (Table 3). S252F-KI and S248F-KI acquired the same amino acid change in the α4-nAChR subunit, but the ES cell lines used for the generation of S252F-KI and S248F-KI were different; J1 and W9.5 ES cells, respectively (Klaassen et al., 2006;Teper et al., 2007). One KI mouse model harbouring the mutant Chrna4, corresponding to the human insL-mutant CRHNA4, was also generated using J1 ES cells, named insL-KI, (Klaassen et al., 2006) (Table 3).   (Klaassen et al., 2006). Both S252F-KI and insL-KI showed important pharmacological features. These two KI mice displayed enhanced sensitivity to nicotine application, because their nicotine-induced seizures showed shorter latencies to seizure onset and longer durations than the wild type (Klaassen et al., 2006).
The slice patch-clamp method demonstrated that nicotine application did not affect the excitatory synaptic current (EPSC) but markedly activated the inhibitory postsynaptic current (IPSC) in layers II/III of the cortex. In contrast, spontaneous EPSC and IPSC were not affected. These results suggest that the enhanced GABAergic inhibition induced by S280F-mutant and insL-mutant α4β2-nAChRs play important roles in ADSHE epileptogenesis/ictogenesis with S280F and insL mutations (Klaassen et al., 2006).

| V287L mutant models
Three genetic animal models harbouring mutant Chrnb2, corresponding to the human V287L-mutant CRHNB2, have been generated (Gullo et al., 2014;Manfredi et al., 2009;Shiba et al., 2015;Xu et al., 2011) (Table 3). One type of transgenic mouse, named V287L-TG, was developed employing a tetracyclinecontrolled promoter, which allowed the researchers to silence the mutated gene in a reversible fashion (TET-OFF system) (Manfredi et al., 2009). The frequency of spontaneous interictal/ictal discharges was increased in the V287L-mutant in a gene expressiondependent manner (10-fold increased β2-nAChR compared to wild type) (Manfredi et al., 2009), but the spontaneous seizure frequency during non-REM sleep was higher than during wakefulness (Manfredi et al., 2009) overexpression of the V287L-mutant β2-nAChR probably contributes to the epileptogenesis of ADSHE, although this is not related to ictogenesis (Table 3) (Manfredi et al., 2009 Xu et al., 2011). The behaviour of V287L-KI mice was generally normal (O'neill et al., 2013;Xu et al., 2011), but abnormalities such as anxiety and natural reward (nicotine behavioural addiction: determination using wheel-running activity), associated with the α4β2-nAChR of V287L-KI, were detected (Table 3) (Xu et al., 2011).
Transgenic rats carrying V286L-TG mutant Chrnb2, corresponding to the human V287L-mutant CRHNB2, display EEG-sensitive NPA (45%) without other ADSHE seizures (NPD and ENW), although they also exhibited hypersensitivity to nicotine-induced seizures (Table 3) (Shiba et al., 2015). Therefore, the face validities of V287L-KI, V287L-TG and V286L-TG were also partly verified, but these imply limited face validities, because the outcomes of the verification study did not show an agreement between the model and clinical observation. However, V287L-TG indicated the important pathomechanism of ADSHE with V287L-mutation, because the V287L-mutant β2-nAChR subunit possibly contributes to the generation of abnormal/epileptic formation of neuronal circuits and/or long-lasting alterations in network assembly during the developing brain.
Carbamazepine was effective at controlling ADSHE seizures in all patients with V287L-mutation , whereas the seizure frequency of V287L-TG could not be reduced by carbamazepine (Manfredi et al., 2009). Contrary to the poor response to carbamazepine in V287L-TG, neuronal firing of primary cultured neocortical neurons of V287L-TG using a multi-electrode array revealed that the inhibitory effects of carbamazepine on neuronal firing of V287L-TG was more dominant than that of the wild type (Gullo et al., 2014). Therefore, the predicted pharmacological validity of V287L-TG was also partly verified, but this implies limited predictive validity, because carbamazepine, which is a first-line anticonvulsant for the treatment of patients with ADSHE with V287L-mutation , suppresses neuronal activity (Gullo et al., 2014) but cannot decrease seizure frequency in V287L-TG (Manfredi et al., 2009).

| S284L mutant models
Two transgenic rat models harbouring mutant Chrna4, corresponding to human S284L-mutant CRHNA4, named S284L-TG and S286L-TG, were generated (  (Fukuyama et al., 2020b;Zhu et al., 2008). Although the S284L-TG and S286L-TG phenotypes fulfil the face validity, the frequency of ADSHE seizures in these models is lower than that in untreated ADNFLE patients (once a week seizure frequency) (Fukuyama et al., 2020b;Zhu et al., 2008). In S286L-TG, clustering is rare, but seizures are observed multiple times a day (Fukuyama et al., 2020a). Additionally, the onset of interictal and ictal discharges occurred at 6 and  (Zhu et al., 2008). S284L-TG also represented the epileptogenic functional shifts, higher basal glutamate release (Zhu et al., 2008), increased NKCC1 and decreased KCC1/KCC2 expression, compared to the wild type (Yamada et al., 2013). Based on these functional abnormalities during the critical period for ADSHE onset in S284L-TG, S284L-mutant α4β2-nAChR or its induced GABAergic disinhibition lead to epileptogenesis. Indeed, chronic furosemide (NKCC1 inhibitor) administration from 4 to 8 weeks of age prevented ADSHE onset in 67% of S284L-TG rats (Yamada et al., 2013). In contrast, the remission rate with chronic furosemide administration after ADSHE onset (between 8 and 10 weeks of age) was only 37% (Yamada et al., 2013). In spite of these efforts, whether the increased frontal extracellular glutamate level in S284L-TG was a functional abnormality associated with epileptogenesis/ictogenesis or a result of the acquisition of epileptogenesis/ictogenesis could not be clarified. Enhanced glutamatergic transmission in the thalamo-subthalamic pathway, from the MoTN to the subthalamic nucleus (STN), was also observed in S286L-TG (Figure 1) (Fukuyama et al., 2020b). Therefore, hyperactivation of glutamatergic transmission in the thalamo-subthalamic pathway contributes to the generation of EEG-insensitive tonic/dystonic or hyperkinetic behaviour (Klaassen et al., 2006;O'neill et al., 2013;Xu et al., 2011). This hyperactivation of the thalamo-subthalamic pathway suggests that hyperactivation of the MoTN, induced by ADSHE-mutant nAChRs, probably propagates predominantly to the basal ganglia rather than the frontal cortex at least, during NPD ( Figure 1) (Fukuyama et al., 2020b). In spite of these efforts, increased basal extracellular gluta- However, it can be reasonably interpreted by other functional abnormalities of the tripartite synaptic transmission (Figure 1) (K. K. Fukuyama, Fukuzawa, Okubo et al., 2020;. After ADSHE onset, connexin43 (Cx43) in the plasma membrane (mCx43) was upregulated in both the thalamus and frontal cortex, whereas before onset of interictal and ictal discharges (4 weeks of age), mCx43 expression in the frontal cortex was not upregulated (K. K. Fukuyama, Fukuzawa, Okubo et al., 2020;.
Cx43 is the most widely and predominantly expressed connexin isoform in the brain, including astrocytes (Okada et al., 2021;. Six connexin units assemble to form homomeric or heteromeric connexons. Two connexons in two neighbouring cells form a gap junction, which contributes to the cytoplasmto-cytoplasm communication of biochemical and ionic mobilisation between adjacent cells, leading to the regulation of ionic and several other types of homeostasis, including regulation of intracellular Ca 2+ mobilisation and K + buffering (Okada et al., 2021;. Single connexons contribute to the chemical connection between intra-and extracellular spaces as a hemichannel (Okada et al., 2021;. During the resting stage, astrocytes are characterised by a high level of gap-junctional communication but low hemichannel permeability (Okada et al., 2021;. In contrast, pathological hyperactivated conditions, such as depolarisation, ischaemia, specific cation mobilisation and phosphorylation, generate persistent hemichannel opening, resulting in the persistent astroglial non-exocytotic release of excitatory glutamate, D-serine, ATP, kynurenine metabolites and eicosanoids, which leads to the disruption of several homeostasis systems (Okada et al., 2021;.
Upregulated mCx43 in the thalamus and frontal cortex of S286L-TG has already been shown to be weakly but functionally activated during the interictal stage (K. K. Fukuyama, Fukuzawa, Okubo et al., 2020;, resulting in increased basal L-glutamate release in the frontal cortex of both S284L-TG and S286L-TG (Figures 1 and 2) (Fukuyama et al., 2020b;Zhu et al., 2008). The lack of change in glutamate release during the transition from wakefulness to non-REM sleep of S284L-TG (Zhu et al., 2008)  The electrophysiological inputs in the thalamus, including RTN, MoTN and MDTN, activate astroglial hemichannel activities, leading to enhancement of thalamo-subthalamic and thalamocortical glutamatergic transmission (Okada, 2019;Okada, Fukuyama, Kawano, et al., 2019;Okada, Fukuyama, Nakano, & Ueda, 2019;Okada, Fukuyama, Okubo, et al., 2019;K. Fukuyama, Fukuzawa, Okubo et al., 2020;Fukuyama et al., 2020a;Fukuyama et al., 2020b;. Enhanced thalamo-subthalamic glutamatergic transmission produces an imbalance in the transmission of the basal ganglia (Fukuyama et al., 2020b). The integration of enhanced thalamocortical inputs (primarily through the MoTN-M2C pathway), on electrophysiology, seem to generate an epileptic discharge (Fukuyama et al., 2020b). In fact, the accumulation of repetitive excitatory thalamocortical inputs reaches the threshold of astroglial hemichannel activation, resulting in the generation of epileptic bursts in M2C (Figures 1 and 2). The OFC is considered one of the dominant focus regions (Provini et al., 1999), although the OFC stimulation induced by intrathalamic GABAergic disinhibition was weaker than that of M2C (Fukuyama et al., 2020a;. Contrary to the interictal stage, the propagation of sustained/repetitive hyperexcitability in the MDTN or OFC issue, because before the onset of interictal/ictal discharges (4 weeks of age), the expression of mCx43 in S286L-TG is almost equal to that of the wild type .
ADSHE onset in S284L-TG can be prevented by chronic administration of the NKCC1-inhibitor furosemide (Yamada et al., 2013). This experiment was based on the findings that upregulation of NKCC1 after ADSHE onset in S284L-TG, which was not observed before ADSHE onset (Yamada et al., 2013). However, the pharmacodynamic profile of furosemide suggests the presence of other pathomechanisms, because furosemide inhibits mitogen-activated protein kinase/extracellular signal-regulated kinase (MAPK/ERK) signalling (Panet et al., 2006). MAPK/ERK signalling regulates protein phosphorylation and cell functions, such as proliferation, division, differentiation, survival and apoptosis Okada et al., 2021;Okada, Kawano, et al., 2020). Transport to the plasma membrane, intracellular communication and degradation of Cx43 are regulated by post-transcriptional processes Ribeiro-Rodrigues et al., 2017). Activated MAPK/ERK and PI3K/Akt signalling increases the expression of mCx43 without affecting mRNA (Cushing et al., 2005;Lin et al., 2013;. Indeed, before ADSHE onset, the MAPK/ERK signalling of S286L-TG was upregulated, whereas Akt signalling was almost equal to that of wild type (Fukuyama & F I G U R E 2 Scheme of proposed hypothesis of age-dependent and event-related pathomechanism/epileptogenesis of ADSHE of S286L-TG. Proposed hypothesis of age-dependent and event (sleep and seizure)-related pathomechanism/epileptogenesis associated with connexin43 and S286L-mutant α4β2-nAChR of S286L-TG. Before ADSHE onset (until 4 weeks of age), loss-of-function S286L-mutant α4β2-nAChRs generate upregulation of ERK signalling and intrathalamic GABAergic disinhibition, resulting in weakly enhanced glutamatergic transmission in the thalamocortical and thalamic hyperdirect pathways. During 4 and 8 weeks of age (critical period for interictal discharge onset), upregulated ERK signalling gradually increases expression of connexin43 in the astroglial plasma membrane. The upregulated astroglial connexin43 hemichannels are activated by both physiological sleep spindle and pathological interictal discharges. At the critical ADSHE onset period (8 weeks of age), the combination of accumulating physiological and pathological bursts and upregulated connexin43 hemichannels leads to enhanced tripartite synaptic transmission, resulting in the generation of ADSHE ictal discharges in the regions where expression of α4β2-nAChRs is predominant . Furthermore, sub-chronic nicotine administration suppressed mCx43 and ERK signalling but enhanced Akt signalling in wild-type cells, whereas sub-chronic nicotine administration enhanced Akt signalling but did not affect mCx43 expression or Erk signalling in S286L-TG . These results indicate that loss-of-function of the ADNFLE-mutant α4β2-nAChR also possibly has loss-of-function of suppressive effects on ERK signalling, leading to upregulation of mCx43.
The increased basal glutamate release of S286L-TG compared to the wild type is also important as the pathomechanism of ADSHE, according to the imbalance hypothesis of epilepsy . During the resting stage, astroglial gap junctions act as permeable intracellular molecules via functionally opening probability, whereas astroglial hemichannel exhibits low opening probability (Jeanson et al., 2015;Liu et al., 2016). Based on these experimental findings, we believe that astroglial hemichannels do not contribute to gliotransmitter release during the resting state due to their low opening probability (K. K. Fukuyama, Fukuzawa, Okubo et al., 2020;Fukuyama, Ueda et al., 2020;Jeanson et al., 2015;Liu et al., 2016). However, currently, we speculate that even in the low opening probability, the upregulation of mCx43 is probably involved in the increased basal release of glutamate in S286L-TG during the interictal state, because astroglial hemichannel inhibitors unexpectedly decreased the increased basal L-glutamate release of S286L-TG, without affecting that of wild type (K. K. Fukuyama, Fukuzawa, Okubo et al., 2020). Therefore, it is undeniable that the increased basal release of glutamate in S286L-TG is provided by glutamate released through astroglial hemichannels.
Previous studies have found that neurotransmitter release requires extracellular K + concentrations of greater than 25 mM, whereas astrocytes require a concentration greater than 100 mM (K. Kawata et al., 1999). Physiological neural activity increases extracellular K + levels up to 1 mM, whereas epileptic discharges increase extracellular K + levels to around 10-12 mM (Carmignoto & Haydon, 2012). Furthermore, mCx43 expression in primary cultured astrocytes is increased by exposure to sustained K + levels higher than 10 mM for several hours . Considering that activated hemichannels release K + into the extracellular space, activated hemichannels induced by eleva- These channels are also involved in memory formation/consolidation, whereas several preclinical behavioural studies have reported that both impaired and hyperactivated astroglial functions contribute to cognitive disturbance, such as loss of short-term spatial memory and fear memory consolidation (He et al., 2020). Interestingly, a post mortem study demonstrated an upregulation of Cx43 in the frontal cortex of patients with autism compared to healthy subjects (Fatemi et al., 2008). Therefore, upregulated mCx43 around epileptic foci or neural circuits for the propaga-  (Takaku & Sango, 2020).
It has been reported that zonisamide affects various transmitter regulation systems, including the candidate targets for the suppression of mCx43 expression, such as ubiquitin ligase, metabotropic glutamate receptors and carbonic anhydrase [Yamamura et al., 2009;Omura et al., 2013;Fukuyama et al., 2014]). Further research is needed to clarify the specific mechanisms underlying the inhibitory effect of zonisamide on astroglial hemichannel expression.

| Effects of nAChRs on the intracellular signalling pathway
The mechanism through which the loss-of-function S286L-mutant α4β2-nAChR increases mCx43 is one of the fundamental scientific issues associated with ADSHE pathomechanism (Figure 2). The primary functions of the nAChR family are ligand-gated cation channels.
The nAChRs containing the α7 subunit (α7-nAChRs) is less sensitive to ACh (EC 50 is micromolar order) and exhibits rapid desensitisation (in the order of milliseconds), whereas α4β2-nAChR is more sensitive to ACh (EC 50 is sub-micromolar order) (Campling et al., 2013). In addition to this rapid electrophysiological response, nAChRs are also known to affect long-term intracellular signalling through several intracellular signalling pathways (Akaike & Izumi, 2018;Schuller, 2009 Nicotine intake upregulates the expression of several genes (Li et al., 2002), including Akt; however, acute and chronic administration of nicotine upregulate and downregulate phosphorylated ERK, respectively (Brunzell et al., 2003;Valjent et al., 2004). Before the ADSHE onset period (4 weeks of age), the expression of phosphorylated Akt and mCx43 in the thalamus and frontal cortex of S286L-TG and wild type was comparable, whereas the expression of phosphorylated ERK was upregulated in S286L-TG . In contrast, after ADSHE onset (12 weeks of age), mCx43, phosphorylated Akt and ERK expression in S286L-TG were upregulated compared to those in the wild type ( Figure 2) . A fundamental issue is establishing whether upregulated phosphorylated ERK prior to ADSHE onset contributes to epileptogenesis, or whether upregulated mCx43 and phosphorylated Akt play critical roles in ADSHE ictogenesis. It has been demonstrated previously that the propagation of epileptic discharges upregulates PI3K/Akt/ mTOR signalling (Talos et al., 2018). Considering that the chronic administration of furosemide (MAPK/ERK inhibitor) prevents ADSHE onset, the combination of enhanced glutamatergic transmission and upregulated MAPK/ERK signalling in S286L-TG shows that the congenital loss-offunction S284L-mutant α4β2-nAChR is critical for the development of epileptogenesis of S286L-TG. In addition, the upregulation of both mCx43 and phosphorylated Akt induced by these two functional abnormalities is also relevant in primary and/or secondary ADSHE ictogenesis.

| Possible pathogenesis of comorbid cognitive impairment
S284L-TG rats do not show general behaviour and sensorimotor function abnormalities (Zhu et al., 2008), and other tests, such as the water maze, passive avoidance, forced swim and pre-pulse inhibition tests, were also normal in this model. However, the spontaneous locomotor activity and social interaction test did detect a behavioural deficit in S284L-TG (Zhu et al., 2008). These results suggest that S284L-TG may develop autism-like neuro-cognition deficits (Zhu et al., 2008) and similar behavioural abnormalities were observed in V287L-KI (Xu et al., 2011). These behavioural abnormalities of S284L-TG and V287L-KI associated with anxiety or social anxiety are insufficient and non-specific, suggesting the acquisition of autism as a comorbidity of ADSHE. Therefore, more detailed behavioural analyses of S284L-TG/S286L-TG associated with autism should be performed, and it should be also clarified whether the behavioural abnormalities of S284L-TG (autism-like cognitive deficits or anxiety) are due to the loss-of-function mutant α4β2-nAChRs.