Mutations affecting GABAergic signaling in seizures and epilepsy
- First Online:
- Cite this article as:
- Galanopoulou, A.S. Pflugers Arch - Eur J Physiol (2010) 460: 505. doi:10.1007/s00424-010-0816-2
- 502 Views
The causes of epilepsies and epileptic seizures are multifactorial. Genetic predisposition may contribute in certain types of epilepsies and seizures, whether idiopathic or symptomatic of genetic origin. Although these are not very common, they have offered a unique opportunity to investigate the molecular mechanisms underlying epileptogenesis and ictogenesis. Among the implicated gene mutations, a number of GABAA receptor subunit mutations have been recently identified that contribute to several idiopathic epilepsies, febrile seizures, and rarely to certain types of symptomatic epilepsies, like the severe myoclonic epilepsy of infancy. Deletion of GABAA receptor genes has also been linked to Angelman syndrome. Furthermore, mutations of proteins controlling chloride homeostasis, which indirectly defines the functional consequences of GABAA signaling, have been identified. These include the chloride channel 2 (CLCN2) and the potassium chloride cotransporter KCC3. The pathogenic role of CLCN2 mutations has not been clearly demonstrated and may represent either susceptibility genes or, in certain cases, innocuous polymorphisms. KCC3 mutations have been associated with hereditary motor and sensory polyneuropathy with corpus callosum agenesis (Andermann syndrome) that often manifests with epileptic seizures. This review summarizes the recent progress in the genetic linkages of epilepsies and seizures to the above genes and discusses potential pathogenic mechanisms that contribute to the age, sex, and conditional expression of these seizures in carriers of these mutations.
Genetic defects that manifest predominantly as seizures or epilepsies (a.k.a. “idiopathic epilepsies”) are not very common. They are known to occur in approximately 1% only of epilepsy patients. Genotype–phenotype studies and the increasing knowledge on the function of these genes have enriched our understanding of the pathophysiology of epilepsy and seizures. Most of the current linkage studies have revealed associations supporting the polygenic origin of idiopathic generalized epilepsies (IGEs). Genetic mutations seem to predispose but are not usually sufficient to cause epilepsy on their own. Among the implicated genes, are those involved in γ-aminobutyric acid (GABA) signaling, the principal inhibitory mechanism in the central nervous system. In this review, I will summarize the key components and features of the GABA signaling and describe the known channelopathies that perturb GABA signaling, either via direct effects upon GABAA receptors or through indirect effects on chloride transporters that ultimately impair GABAA signaling. Finally, I will also discuss how the functional consequences of these mutations are believed to explain the key features of the respective epileptic syndromes or seizures.
GABA signaling: basics on biology and physiology
Whether through phasic or tonic conductance, GABAA receptors mediate the principal inhibitory function in the brain. Their function, however, changes depending upon the specific conditions that control the chloride gradient [1, 2, 4]. When chloride accumulates intracellularly, GABAA receptor activation has depolarizing effects. This is classically the case in immature neurons, in which the expression or activity of chloride transporters that favor the import of Cl− into the cell, such as the sodium potassium chloride cotransporters (NKCCs), override the effects of chloride exporting proteins, like the potassium chloride cotransporters (KCCs). In every neuron, there is a gradual change in Cl− gradient during development, attributed to a gradual switch from NKCC to a KCC dominant state, which leads to low intracellular Cl− and allows the hyperpolarizing effects of GABAA receptors to manifest. The importance of KCCs in promoting this shift was elegantly demonstrated by Rivera et al. who showed that KCC2 activity is necessary for the appearance of hyperpolarizing GABAA signaling in hippocampal neurons . KCC2 has been one of the most studied chloride cotransporters in regard to its role in GABAA receptor function and seizures, and its importance in both GABAA signaling and brain development has been well documented [1, 2, 6–9]. However, KCC2 mutations have not yet been identified in patients with epilepsy or seizures. It is possible that disruption of KCC2 may result in early mortality, similar to what has been observed in mice , precluding the identification of KCC2 mutation carriers. In contrast, KCCs that may have more restricted expression or may not be as essential for survival may result in an identifiable phenotype. In this review, reference will be made to KCC3 mutations that have been linked with seizure disorders.
GABAB receptors are metabotropic receptors, consisting of seven transmembrane domains. They are localized at pre-, post-, or extrasynaptic sites as functional heterodimers. Homodimers of R1 or R2 GABAB receptor subunits have been described with distinct intracellular localization and can under certain conditions be directed to the plasma membrane [11–13]. Although there is no evidence currently that homodimers are functionally active, atypical responses to GABAB receptor agonists have been described in the GABAB R2 knockout mouse . By forming heterodimers, they can inhibit calcium channels (presynaptic effects) and block as a result the presynaptic release of neurotransmitters, whether these may be inhibitory (i.e. GABA) or excitatory (i.e. glutamate). In the postsynaptic, and possibly presynaptic, compartment they may also activate inward rectifying potassium channels (Kir), hyperpolarizing the neurons, and generating inhibitory postsynaptic potentials (IPSCs). Furthermore, they may inhibit adenylate cyclase postsynaptically influencing cAMP-dependent cell signaling .
Currently, there has been significant progress in genetic linkage studies associating seizure or epilepsy disorders with specific components of the GABA signaling pathways. In this review, I will focus on the GABA channelopathies that are linked to increased susceptibility to human epilepsies and seizures. Specifically, I will describe the human mutations of GABAA receptors, as well as on mutations of chloride transporters and channels that perturb chloride homeostasis and disrupt GABAA signaling, leading to epileptogenesis.
Epilepsy syndromes and seizures associated with GABAA receptor mutations
Childhood absence epilepsy
GABAA receptor mutations in epilepsy and seizures
Seizure type / epilepsy
Type of mutation
Site of mutation
Functional effect of mutation
Autosomal dominant JME
Low amplitude GABA currents; reduced surface expression; increased GABA EC50
Low amplitude GABA currents; impaired trafficking?
GABRA1 (975delC; S326fs328X)
Frameshift mutation, Premature termination
Nonsense-mediated mRNA decay; degradation in ER
No change in GABAA current
GABRB3 (−897T/C polymorphism)
Decreased transcription (impaired N-Oct-3 binding)
GABRB3 (P11S; S15F; G32R)
CAE + febrile seizures
Decreased surface expression; ER retention; Altered benzodiazepine / GABA binding?
(IVS6 + 2 T>G)
Exon 6 skipping, Premature termination
Exon skipping, premature termination, ER retention and degradation vs nonfunctional protein.
M2–M3 linker (extracellular)
Decreased GABAA current amplitude; Intact DZ potentiation
M3–M4 linker (intracellular)
GABRD (E177A) (R220C)
Low amplitude GABA currents; impaired trafficking?
Also seen in controls
IGE, ADEFS+, febrile seizures
GABRE (Y38S; E52K)
Also present in controls
M3-M4 linker (intracellular)
Increased fast phase desensitization, reduced sensitivity to diazepam
Deletion, maternally inherited
Decreased expression of involved genes (may include GABRB3, GABRA5, GABRG3)
GABRA1 (975delC; S326fs328X)
Screening of 98 patients with IGEs (38 had CAE) identified a deletion of a single base pair at position 975 (975delC) of the α1 subunit, which was not seen in three unaffected relatives . This deletion causes a frameshift and premature termination within the third transmembrane domain (M3) of the α1 subunit . Transfection of wild type (α1-WT) or α1975delC constructs labeled with enhanced green fluorescent protein (EGFP) in human embryonic cells (HEK293) resulted in the retention of the truncated α1975delC subunit within the cytoplasm and endoplasmic reticulum (ER) . The inability of the truncated protein to be inserted into the plasma membrane is probably attributed to the propensity of its mRNA to be degraded through the nonsense-mediated mRNA degradation pathway (NMD), as well as to the protein degradation through the ER-associated degradation pathway (ERAD) . Concurring with the absence of this mutant from the cell membrane, no GABAA currents were recorded in these α1975delCβ2γ2 transfected cells . Coexpression of α1975delC with α1-WT did not reveal any dominant negative effect .
A missense mutation of the extracellular segment of this subunit was found in a patient with CAE with atonic seizures but without any evidence of pathological alteration of GABAA currents .
GABRB3 (−897 T/C polymorphism)
Among 45 patients with CAE, 28 manifested a polymorphism in the promoter of the exon 1a of the GABRB3 subunit (−897 T/C; haplotype 2) that reduced the affinity for binding of the neuron-specific transcriptional activator N-Oct-3 . Although haplotype two was not restricted to CAE patients (also present in 10/90 control samples), it increased the risk for CAE. Functionally, this haplotype reduces the expression of the GABRB3 subunit.
GABRB3 (P11S; A15F; G32R)
Screening of 48 families of probands with remitting pyknoleptic absence produced associations with three missense mutations at positions 11, 15 (exon 1a), and 32 (exon 2) of the β3 subunit in members of four families . The probands had absence seizures that could be associated with eyelid myoclonias (P11S, S15F) or grand mal seizures (S15F) or atonic seizures and vomiting (P11S). Incidental presence of the G32R mutation was also found in two siblings of the proband who had 3–4 or 5–6 Hz spike-wave complexes in their EEG with or without prior history of febrile seizures . Asymptomatic carriers were also found among relatives of the proband with the S15F mutation.
In vitro translation of the mutated GABRB3 yielded proteins with excessive glycosylation and reduced amplitudes and densities of GABAA currents in transfected cells, suggesting a decreased number of functional GABAA receptors at the plasma membrane . Whether this is due to impaired trafficking or hypofunctional GABAA receptor complexes remains to be shown.
Febrile seizures and autosomal dominant epilepsy with febrile seizure plus syndromes
Seizures in the context of a febrile illness (i.e. febrile seizures) are very common in the infantile and childhood years, occurring in approximately 2–5% of children. Febrile seizures do not typically raise the risk of subsequent epilepsy, unless they occur as focal, prolonged or clusters of seizures, or when they are associated with neurodevelopmental abnormalities or a family history of epilepsy . Febrile seizures are more likely to recur in children with a family history of febrile seizures, although the inheritance is thought to be usually multigenic . A subgroup of these families demonstrates an autosomal dominant mode of inheritance [21, 22]. Reports of co-segregation of patients with different types of idiopathic generalized or focal epilepsies and/or history of febrile seizures within the same family pedigrees emerged over the past decade, bringing onto light a number of susceptibility genes. Initially, the term Generalized Epilepsy with Febrile Seizures Plus syndrome was given as a descriptor . However the term Autosomal Dominant Epilepsy with Febrile Seizure Plus Syndromes has also prevailed, as it was recognized that both focal onset and generalized afebrile seizures could manifest within these pedigrees, and the trait was inherited in an autosomal dominant mode. Although the most common gene mutations are sodium channelopathies, a number of GABAA receptor mutations have also been implicated, as described below.
Mutations linked with CAE and febrile seizures syndrome
One of the first associations of GABRG2 mutations and IGEs was reported by Wallace et al.  in a large pedigree of patients with CAE and febrile seizures with or without other types of seizures. Most of the affected members carried the γ2R43Q mutation. Asymptomatic carriers were also found in older generations of this pedigree. In parallel, the same pedigree also included members who had either EEG abnormalities or other seizure types without carrying the γ2R43Q mutation. The involvement of compensatory factors that suppress the expression of seizures in selected subjects is therefore suggested, whereas the influence of genetic or epigenetic factors modifying the phenotype cannot be excluded.
A recent knockin mouse constructed to express γ2R43Q provided a proof of principle that heterozygous expression of this mutation is sufficient to induce behavioral arrests with bursts of spike-wave complexes (5–8 Hz) in the majority of mice . These seizures first appeared around postnatal day 20 (equivalent to human childhood) and were suppressed by ethosuximide, similar to absence seizures . The genetic background modified the probability to express these seizures, explaining the variable expression of seizures among carriers .
Several studies have addressed the effects of γ2R43Q mutation not only on the basic electrophysiological properties of the GABAA receptor complexes, but also upon benzodiazepine binding, receptor trafficking, and complex formation. A uniform finding is the decreased trafficking of the γ2R43Q subunit to the cell surface, due to its retention and possible degradation within the ER [25–28]. Such studies utilized coexpression with the α1β2 or α1β3 subunits. In cell-culture experiments using heterologous transfections, a dominant negative effect has also been reported. The γ2R43Q mutation reduced the surface expression of the β2 subunit , α1β2 complexes , a5 subunits , and re-routed the expression of γ2 or pentameric α1β2γ2S or α1β2γ2SR43Q receptor complexes to perinuclear and ER sites . However in embryonic cultures from γ2R43Q knockin mice, no changes in the trafficking and localization of α1 subunits were seen , raising the possibility that in vivo conditions may modify the dynamics of subunit interactions and trafficking. These may include post-translational modifications, compensatory expression of other subunits, or regulatory factors. Indeed, even in the same model system of γ2R43Q knockin mouse, regional differences in the effects of GABAA receptors were observed. The cortical pyramidal (layers II/III) neurons of γ2R43Q knockin mice manifested the greatest impairment in GABAA neurotransmission, whereas GABAA function in the thalamic nuclei was unaffected . This generated the hypothesis that cortical hyperexcitability may underlie the pathogenesis of absence seizures in these mice. In agreement, increased cortical excitability was also seen in human patients with the γ2R43Q mutation, using transcranial magnetic stimulation . Despite the significant reduction in surface availability of γ2R43Q containing GABAA receptor complexes, a small number of assembled complexes eventually escape intracellular degradation forming complexes. As a result of the decreased number of functional membrane GABAA receptors, IPSC amplitudes are smaller in most studies.
R43Q is located at the γ2/β subunit interface, in the proximity of the first high-affinity benzodiazepine binding site . Although it does not seem to be directly involved in benzodiazepine or GABA binding, it may indirectly affect them, in certain model systems, possibly through conformational changes [24, 28, 29, 32, 33]. However, this has not been supported by other studies [26, 27]. Interestingly, in humans, decreased benzodiazepine receptor binding has been found with (11C)-flumazenil positron emission tomography in insular and anterior cingulated cortex , although this may probably reflect the reduced overall membrane expression of GABAA receptors.
Apart from altering synaptic physiology, the γ2R43Q mutation has also been proposed to influence brain plasticity . Conditional suppression of this allele in knockin mice from conception till PN20 decreases the susceptibility to pentylenetetrazole seizures in adulthood, when compared to mice with life-long expression of γ2R43Q allele. A plausible proposed explanation is that disruption of the physiological effects of GABAA receptors during the sensitive development fetal and neonatal/infantile life may compromise the developmental processes that are crucial for normal brain development and facilitate, possibly, the development of epilepsy [1, 4, 35, 36].
A splice donor site mutation in intron six of the GABRG2 subunit was identified in a family with CAE and febrile seizures . This is predicted to cause the skipping of exon 6, creation of a premature stop codon at the junction of exons 5 and 7, and nonfunctional truncation upstream of the transmembrane domain M1.
Mutations linked with ADEFS plus syndrome
A missense mutation K289M, at the linker of M2 and M3 domains, was identified in members of a family with generalized tonic–clonic epilepsy and febrile seizures, who were therefore classified as autosomal dominant epilepsy with febrile seizure plus syndrome (ADEFS+) . The K289M mutation causes significant reduction in the amplitude of the GABAA currents and shortening of the receptor open time, whereas benzodiazepine sensitivity is retained [26, 38–40]. Interestingly, phenobarbital potentiates α1β2γ2K289M receptors to a greater extent than wild-type receptors improving therefore GABAA inhibition .
In a French family with ADEFS+ phenotype a missense mutation Q351X was identified . Heterologous expression of GABAA receptors with this subunit did not produce any functional chloride channels due to intracellular retention of the mutated subunits. This mutation is predicted to cause premature termination between M3 and M4 domains.
The W390X nonsense mutation causes premature termination between M3 and M4 domains. It was identified in a Chinese family with several members with febrile seizures with or without other types of seizures .
A missense mutation at the extracellular segment has been linked with febrile seizures . Functionally, the receptors had increased desensitization and reduced sensitivity to diazepam.
GABRD (E177A; R220C)
GABRD containing receptors are typically extrasynaptic and mediate tonic GABAA inhibition after being activated by ambient GABA . Missense mutations E177A and R220C were identified in families with ADEFS+ or febrile seizures . The electrophysiological effect of the E177A mutation is the reduction of the peak current under saturated doses of GABA without changing binding affinity; decreased open times and reduced cellular expression have been proposed as possible causes for this deficit .
Screening of patients with ADEFS+, IGEs or febrile seizures for mutations or polymorphisms at the known GABAA receptor subunit genes revealed rare additional missense mutations at the GABRP (V10M) and GABRE (Y38S, E52K, G66S), GABRA4 (T320A) genes. Rare occurrence in control patients was also reported for the GABRA4 (T320A) and GABRE (Y38S, E52K) mutations . The functional and pathogenic significance of these mutations is currently unclear.
Severe myoclonic epilepsy of infancy (Dravet syndrome; SMEI)
SMEI or Dravet syndrome manifests in the first year of life with clusters or prolonged febrile seizures (clonic, tonic, and generalized tonic–clonic), which can be difficult to suppress and frequent evolution to other epilepsies and developmental problems . The known GABAA receptor-related mutations that have been linked so far with this condition are nonsense mutations, leading to premature termination of the γ2 (Q351X or Q40X) subunits; ER retention reduced expression of these subunits in the cell surface.
An isolated case of a patient with SMEI within an Austrian family with ADEFS+ linked to a Q351X nonsense mutation of the γ2 subunit was reported . Other carriers of the same mutation had milder seizure types.
This nonsense mutation has been reported in patients with SMEI, causing premature termination, ER retention, and reduction in the surface expression of the γ2 subunit . Asymptomatic carriers were also found within this family.
Autosomal dominant juvenile myoclonic epilepsy
Juvenile myoclonic epilepsy is an autosomal dominant form of IGE that typically begins when the child approaches the teenage years. The classical seizure type is the myoclonic jerks upon awakening, whereas grand mal seizures (almost in all patients) and absence seizures (in a third of patients) are also very common . JME can appear within the spectrum of ADEFS+ and therefore can be linked to the mutations discussed above. However, two specific GABAA receptor mutations have been associated specifically with JME.
The A322D mutation is located within the M3 transmembrane domain, close to a region that is thought to undergo a conformational change upon binding of agonists or allosteric modulators of GABAA receptors [51–53]. Heterologous expression of the missense mutation α1A322D in HEK293 cells produces hypofunctional GABAA receptor channels (α1A322Dβ2γ2). In transfection experiments in HEK293T cells, differences in the properties and kinetics of the assembled channels were found under conditions favoring the formation of wild-type, heterozygous, or homozygous GABAA receptors. Heterozygous channels had reduced current amplitudes yet similar activation, desensitization, and deactivation kinetics with the wild-type channels. In contrast, homozygous channels had accelerated deactivation and resensitization kinetics, significantly reduced open times and affinity for GABA, as well as very low current amplitudes [54–57]. Reduced whole-cell and cell-surface expression of the mutated channels has been proposed to explain the reduced amplitudes .
Heterozygous or homozygous δR220H mutations were found in a family with JME, as well as sporadically in control patients . This mutation is present at the extracellular domain of the GABAA receptors and results in low amplitude GABAA currents, due to reduced open times and, to a lesser extent, decreased expression of the receptor onto the cell membrane . However, the mutation effect upon the membrane expression of the receptors was rescued by the presence of a wild-type allele .
The Angelman syndrome appears in the first year of life, with an estimated frequency of one in 15,000 births . Infants with normal development in the first months of life manifest developmental delay, followed by more permanent intellectual and developmental disabilities, including speech, gait, and balance impairments, hypotonia, epileptic seizures (infantile spasms, myoclonic seizures and absences, tonic or tonic–clonic seizures, and partial seizures), episodes of arm flapping, tongue thrusting, and head drops, yet they have a happy appearance and paroxysms of laughter [50, 58]. Their EEGs often demonstrate slow, high amplitude spike- and slow-wave complexes (SWDs) or dysrhythmia (rhythmic slowing frontal or occipital or diffuse) . In the majority of cases, a maternally inherited deletion of 15q11-13 locus is identified, which includes the UBE3A gene (ubiquitin-protein ligase E3A), as well as three GABAA receptor subunit genes: the GABRB3, GABRA5, and GABRG3 [60–62]. Minassian et al. correlated genotype with phenotype in patients with Angelman syndrome and proposed that patients with 15q11-13 deletion, that included the deletion of GABR genes in addition to UBE3A, follow a more severe course than patients with UBE3A mutations . Patients with 15q11-13 deletion often had flexor or extensor epileptic spasms in addition to atypical absences or myoclonus . In contrast, patients with UBE3A mutations have only a mild phenotype.
Although the GABRB3−/− knockout mice have additional characteristics, related to the GABA receptor deficit, such as cleft palate and decreased viability, the survivors develop the neurodevelopmental deficits and epilepsy phenotype of Angelman syndrome: neurodevelopmental impairment, stereotypical behaviors, such as running in circles, hyperactivity, impaired motor development, abnormal electrographic patterns, and seizures . Seizures include myoclonic jerks and seizures, head drops, opisthotonus and tail arching, sudden drops, and wild running episodes. The brains of GABRB3−/− knockout mice also showed significant widespread reductions in the binding affinity for both direct and benzodiazepine-like GABAAergic agonists. The authors proposed that the deficit in β3 may also decrease other non-β3-containing, zolpidem-insensitive, GABAA receptors . Decreased sensitivity to certain anesthetics, like midazolam and etomidate, has been described in the GABRB3−/− knockout mice . An interesting sex-specific pattern of epigenetic modification of the β3 gene was reported in the β3+/− heterozygotes, where decreased β3 expression was seen in all females, regardless of the parental origin of the affected allele, whereas males that demonstrated low β3 levels only in the affected allele were of maternal origin .
Clinical and pathophysiologic implications of GABAA receptor mutations in epilepsies
Pathophysiological role of GABAA receptors in IGEs
Implications for age and sex-specific features of IGEs
IGEs demonstrate a classical, defining for each syndrome, age-specific pattern of expression. This may correspond potentially to the ages when the presence of a pathogenic mutation of the specific GABAA receptor subunits directly or indirectly (i.e. through dominant negative effects or interactions with other subunits or pathways) weakens the intrinsic defense mechanisms that control seizures. Three of the GABAA receptor subunits discussed here, α1, β3, and γ2, exhibit distinct developmental patterns of expression, based on studies in rodents, yet all three have been implicated in CAE. In the corticothalamic network, α1 and γ2 increase during the infantile and juvenile periods, whereas β3 declines [19, 70, 71]. It is therefore also possible that the ages of vulnerability to childhood absence seizures may coincide with the developmental periods when rapid changes occur in the expression of these subunits. As a result of these dynamic changes in gene expression, the brain may not be able to fully compensate for deficiencies in the affected subunit, by increasing the availability of other subunits. Resolution of absence seizures may then occur either because the biological role of the affected subunit in the corticothalamic network may, normally, diminish at older ages (as postulated for the β3 subunit ) or because increase in other subunits or the wild-type alleles may eventually compensate.
GABAA receptors are also key regulators of seizure susceptibility, whether these are generalized or focal, by influencing the activity of subcortical centers, like the substantia nigra, that control seizure propagation [72–74]. The immaturity of the GABA-sensitive seizure controlling subcortical networks, early in life, is an important factor for the age-related vulnerability to certain seizure types, like absence seizures. The presence of the above GABAA receptor mutations may further deteriorate their ability to curtail seizures, leading to epilepsy. Further studies are needed to elucidate the impact of these mutations in the function of these subcortical structures.
IGEs may also show preferential expression in one sex. Two thirds of CAE patients are girls whereas a trend for higher incidence in boys is seen with ADEFS+ and grand mal on awakening . It is therefore not very surprising that GABAA receptors, well-known for its age- and sex-specific patterns in expression and function, are involved in the susceptibility to these syndromes [70, 76–80]. As the signature of IGEs is the SWDs on the EEG (Fig. 3), changes in GABAA receptor expression and function within the corticothalamic network, the generator of SWDs, have been a focus of attention. In a study utilizing Long–Evans rats, lower developmental expression of the α1 and γ2 subunits was seen in the cortex and thalamus of female rats, that correlated with higher frequency of SWDs in females following treatments that induce SWDs in these rats, i.e. perinatal administration of the cholesterol synthesis inhibitor AY9944 [70, 81]. Moreover, AY9944 reduced the expression of these subunits in the female, but not the male thalamus . Such findings are in support of the hypothesis that decreased availability of these two subunits, either via loss-of-function mutations or by impaired trafficking, may increase the susceptibility to CAE, and that this vulnerability may be greatest in girls. However, the rarity of these mutations does not allow definite conclusions yet as to their contribution in the sex-specific prevalence of these disorders.
An alternative mechanism underlying the sex-specific prevalence of epilepsies linked to β3 mutations was presented by Liljelund et al. , who reported sex-specific imprinting of the β3 subunit in heterozygote GABRB3+/− mice. In female heterozygotes, β3 subunit expression was significantly reduced whether the affected gene was inherited by the father or the mother. In contrast, in male heterozygotes, significant reduction was noted only through maternal transmission of the affected gene. Genomic imprinting of the paternal chromosome 15 is known to also occur in human patients with Angelman syndrome. In these cases, however, paternal imprinting explains the preference for transmission through the maternally transmitted affected allele, without contributing to any sex-specific prevalence of the disease [82, 83].
Beyond the immediate consequences of impaired GABAA receptor signaling due to the above mutations, there are wider and long-term implications relating to disruption of GABA-dependent developmental processes that shape the normal and sex-appropriate brain development [1, 4, 9, 76, 84–88]. GABAA signaling controls morphogenesis, neuronal migration and differentiation, as well as synaptogenesis. Early life events that alter GABAA signaling may leave their long-term imprints on the way the brain works and responds to the environment. An elegant proof-of-principle example was offered by Chiu et al. who showed that transient suppression of the expression of the γ2R43Q, early in life, has lasting changes by decreasing susceptibility to seizures in adulthood compared to mice with constant expression of the mutant allele .
Fever, GABAA receptor mutations and seizures
An attractive theory as to the role of fever in unmasking the seizure vulnerability of patients with specific GABAA receptor mutations was proposed by Kang et al., based on a temperature-sensitive impairment in intracellular trafficking . Specifically, a rapid increase in temperature from 37 to 40°C within 10 min impaired trafficking or decreased endocytosis of γ2R43Q subunits, but did not affect subunits with mutations not linked to febrile seizures, like the α1A322D. However, febrile seizures can often occur at the onset of a febrile illness, with minor changes in temperature, whereas in epilepsies like ADEFS+ febrile but also afebrile seizures may occur in the family pedigree. Additional factors need therefore, be invoked to explain susceptibility to febrile seizures and epileptogenesis in such instances. Possible culprits may be immune-related factors, such as cytokines, which increase excitability and excitotoxic injury [90–92]. A genetic background with deficient GABAA inhibition due to the above mutations may therefore, augment the brain excitability due to inflammatory factors and lead to seizures and subsequent epilepsy.
Other genes modulating seizure susceptibility
Chloride channel 2 (CLC-2)
Chloride channel 2 (CLC-2) is a hyperpolarization-activated, inward rectifying Cl− channel that activates at membrane potentials more negative to the reversal potential for Cl− (ECl) [93, 94]. As a result, when Cl− loading occurs through the repetitive activation of hyperpolarizing GABAA receptors, the activation of ClC-2 leads to compensatory efflux of Cl−, maintaining the intracellular Cl− concentrations low. This permits the hyperpolarizing actions of GABAA receptors. The developmental increase in CLC-2 has been proposed as one of the factors that contribute to the gradual decline in ECl in neurons and the appearance of hyperpolarizing GABAA currents [95, 96]. The localization of its gene (CLCN2) within the 3q24 chromosomal region, previously linked to IGEs, prompted the screening of patients with IGEs for mutations in CLCN2. Mutations in CLCN2 have been found in several patients with IGEs, including JME or Juvenile absence epilepsy (JAE) [97, 98] or focal epilepsy . In a study of 112 patients with such epilepsies and 192 control subjects, mutations present in the epilepsy cohort only were found in 1.7% subjects .
CLCN2 (IVS17-3C> T)
This single nucleotide polymorphism (SNP) was found in a small IGE family, including an unaffected parent. However, there is no evidence for a pathogenic role for this mutation, as no change in splicing or expression of CLC-2 was identified .
CLCN2 (R688G, E718D)
The R688G and E718D missense mutations were found in a father and a son who had absence and generalized seizures or frontal lobe epilepsy, respectively . The affected amino acids are located within the carboxy-terminal segment of the CLC. Based on their location, it has been proposed that they interfere with voltage gating . However, this was not confirmed by subsequent studies .
In a different study of three families of patients with epilepsy, three additional mutations were found, that were not present among 4,700 control patients . These were, however, not restricted to patients with epilepsy but were also sporadically present in their asymptomatic relatives, suggesting that they may be susceptibility mutations .
CLCN2 (Gins597; p.M200fsX231)
The p.M200fsX231 mutation leads to premature termination of CLC-2, rendering it nonfunctional. It was detected in a patient with JME, as well as in asymptomatic relatives . Coexpression of the wild type and the M200fsX231 mutant did not reveal any dominant negative effect upon the gating of the co-expressed wild-type CLC-2 channels [99, 100].
Atypical splicing IVS2-14del11, that predicts the deletion of 44 amino acids (del74-117) and skipping of helix B of CLC-2, was found in a patient with IGE, as well as some of his asymptomatic relatives . Such a deletion should result in nonfunctional channels [100, 101]. However, expression of a minigene with the IVS2-14del11 mutation in N2a neuroblastoma cells produced both normally spliced products, as well as mRNAs bearing the expected deletion, suggesting that its pathogenicity may either not be significant or be cell-type selective .
A missense mutation G715E was found in two siblings with JAE or spike-wave activity in the EEG, as well as their father . This mutation occurs at the cytoplasmic carboxy-terminal ending of the protein, between the two cystathionine-β-synthase domains CBS1 and CBS2 [97, 100, 101]. Although initial studies suggested that it may alter the voltage-gating properties of the channel , this was not confirmed . Independent laboratories showed that the G715E mutant of the CLC-2 channel exhibits decreased sensitivity to intracellular adenosine nucleotides, raising the possibility that the mutant allele may impair the activation of the CLC-2 channel under conditions of energy depletion (i.e. low ATP, increased AMP levels) [100, 102]. For example, the binding affinities for ATP or AMP are reduced at least tenfold in the G715E mutant . Furthermore, replacement of intracellular ATP by AMP does not accelerate the kinetics of gate opening in the G715E CLC-2 channel as it does in the wild-type channel . However, it remains to be shown whether such aberrations from normal function are important enough to generate a pathogenic phenotype, at least under conditions of energy depletion.
Overall, the involvement of CLCN2 mutations in idiopathic generalized and/or focal epilepsies is unclear and may represent innocuous mutations. At best, these may be considered as susceptibility gene mutations, rather than pathogenic, especially if one considers that most of these patients with epilepsy carry heterozygous CLCN2 mutations, which do not compromise the functional role of the wild-type CLC-2, at least under normal conditions. Whether this balance can be tipped, leading to epilepsy, in the presence of additional pathogenic or susceptibility mutations, under specific pathological conditions, or during vulnerable developmental periods when the expression of wild-type CLC-2 is still low and unable to fully compensate for haploinsufficiency states due to loss-of-function mutations, still remains to be shown. Of interest, heterozygous loss of CLC-2 had no overt behavioral or morphological consequence in mice, whereas homozygous CLCN2 knockout mice had leukoencephalopathy, blindness, auditory conduction deficits, but no evidence of spontaneous seizures or increased susceptibility to flurothyl-induced seizures .
Potassium chloride cotransporter 3 (KCC3; SLC12A6)
KCC3 (c.2436delG; T813fsX813)
In a cohort of 64 French Canadian patients with HMSN/ACC [110, 111], 70% were homozygous for a truncating mutation of KCC3 that deleted the majority of the carboxy-terminal intracellular segment. Seventeen percent of these patients had infrequent seizures [111, 112] and 8.4% had epileptiform EEGs .
KCC3 (p.Y678LfsX41 / p.F493CfsX48)
A patient carrying two mutations of KCC3 leading to premature termination and truncations following the 6th or 12th transmembrane domains was reported with features of HMSN/ACC and seizures .
KCC3 (p.I301SfsX15 / p.I301SfsX15)
Homozygous truncating mutations of KCC3 at the extracellular linker of the 3rd and 4th transmembrane domain were found in a 5-year-old patient with seizures and HMSN/ACC .
KCC3 (1584_1585delCTinsG; F529fsX532)
This mutation truncates the protein at the 7th transmembrane domain. It co-segregated with the T813fsX813 mutation in a French Canadian patient with HMSN/ACC .
KCC3 (c.3031C->T; p.R1011X)
The R1011X mutation has been identified in several non-French-Canadian patients with HMSN/ACC of Turkish, Italian, African, and Dutch descent and results in the deletion of a small fragment from the carboxy-terminus of the protein [111–113]. Of the patients, 4/8 had seizures or epileptiform EEGs . The truncated protein was nonfunctional when expressed in Xenopus oocytes .
Additional mutations have been reported with HMSN/ACC, including in non-French Canadians, but with no clear co-morbidity with seizures. These include the R675X mutation that truncates KCC3 after the 10th transmembrane domain ; the missense p.G539D mutation in association with the splicing mutation c.1118+1G>A ; the c.619C>T, p.R207C missense mutation ; the E1015X mutation in a Sudanese family . In contrast, no pathogenic mutations of KCC3 were identified in the screening of 16 patients with rolandic seizures and 23 patients with IGE .
In confirmation of the clinical pathogenicity of these truncating mutations, mice homozygous for KCC3 gene knockouts exhibited deficits in locomotion, paired pulse inhibition, and hypomyelination and axonal swelling in the peripheral nerves [109, 112, 118]. Neurodegeneration followed the pathological evidence of cell swelling, suggesting that impaired cell-volume regulation may underlie the cellular degeneration . Although peripheral nervous system pathology was a common finding in the various knockout mice for KCC3, the central nervous system (CNS) was not always affected [109, 112]. When present, it involved the hippocampus but not the corpus callosum, a key site affected in the human phenotype, suggesting that species differences or technical factors that relate to the utilized constructs may influence outcomes . Decreased seizure threshold to flurothyl seizures was reported in the KCC3−/− with CNS involvement . Mechanisms that could underlie this lowered threshold to seizures may primarily relate to the cell-volume dysregulation and cell swelling, which not only can induce neurodegeneration, but also directly increases seizure susceptibility . KCC3 mutations also increase intracellular Cl− and shift the GABAAergic responses to less hyperpolarizing values. However, this effect is expected to be at best modest, as the lack of KCC3 may be compensated by more potent KCCs, like KCC2 .
Chloride transport and genetics of human epilepsies
Type of mutation
Site of mutation
Functional effect of mutation
Loss of function
CLCN2 (IVS2-14del11; del74-117)
Atypical splicing; the expected deletion of helix B not confirmed
Pathogenicity not confirmed
JAE, spike-wave EEG
Pathogenicity of heterozygous mutations not confirmed; Alters AMP-dependent gating.
HMSN/ACC (Andermann syndrome)
KCC3 (c.2436delG; T813fsX813)
Deletion of carboxy-terminal intracellular segment
Loss of function; Impaired cell volume regulation in hypotonic conditions and increase in ECl
KCC3 (p.Y678LfsX41 / p.F493CfsX48)
Truncation at TM7 or after TM12
KCC3 (p.I301SfsX15 / p.I301SfsX15)
Linker of TM3-TM4
KCC3 (1584_1585delCTinsG; F529fsX532)
Truncation at TM7
KCC3 (c.3031C−>T; p.R1011X)
Deletion of carboxy-terminal intracellular segment
What do these mutations reveal about neuronal excitability, ictogenesis, and epileptogenesis? As summarized in Table 1, most GABAA receptor mutations reduce the efficiency of GABAAergic neurotransmission. It is currently unknown whether and when the change to hyperpolarizing GABAA signaling occurs in the human brain and how these tempos are different in brains that have had different prior life experiences (i.e. seizures, stress, fever, etc.). Whether GABAA signaling is depolarizing or hyperpolarizing, reduction of available open GABAA receptors will be expected to reduce shunting inhibition promoting the excitability of a neuron. In regard to ictogenesis, the main question is what the impact will be in a specific network as this will be the determining factor for the type of seizures that will occur. It is interesting that most of the available known mutations have been linked with generalized epilepsies associated with SWD (i.e. CAE, JME) or with hyperexcitability in the setting of a trigger (i.e. fever). Do these indicate that the corticothalamic network is more vulnerable to dysfunction of GABAA signaling or that the mutations linked with febrile seizures or ADEFS+ are simply susceptibility mutations? If the latter, why is the prevalence of other types of seizures and epilepsies, like temporal lobe epilepsy, not as high in these carriers, at least with the association studies that exist so far? Could the lower prevalence of non-SWD seizures reflect a higher degree of redundancy for such isolated GABAA receptor subunit deficits within the networks that generate them, possibly due to local compensatory factors? Or could the early development of IGEs associated with SWD prevent epileptogenesis outside the corticothalamic network, as suggested by Onat’s group . Does the bias in screening patients with strong family history interfere with our ability to unravel the full extent of their involvement in the other types of seizures and epilepsies? Obviously, a seizure susceptibility gene mutation is more likely to be linked with an inherited type of seizure, if we mostly screen patients predisposed to such seizures. A strong argument against this possibility, for the γ2R43Q at least, is the recreation of only the absence seizure phenotype, and not other seizure types, in the γ2R43Q knockin mice of various backgrounds . More research needs to be done to better document the answers to all these questions.
Do the described genetic mutations promote ictogenesis or epileptogenesis? In the constant presence of the original genetically determined cause of deficient GABAAergic transmission, it is hard to firmly state that the ongoing seizures are not simply a result of a constantly lowered seizure threshold due to these mutations. In the classical models of acquired epileptogenesis, we are seeking evidence that either spontaneous seizures can occur in the absence of the original trigger (as in kainic acid induced epileptogenesis) or that seizure threshold decreases with time (as in kindling paradigm). The process however of genetically determined epileptogenesis will have to be viewed as entirely different from acquired epileptogenesis. Genetically determined epileptogenesis may indeed depend more upon the constant presence of causative or contributory genetic deficits that promote excitability of a seizure network and less upon plastic changes. However, the brain, especially during development, is highly plastic and responsive to even the smallest intrinsic or exogenous changes, especially as it pertains to GABA signaling [1, 4, 9, 76, 85, 86, 123]. It is therefore not surprising that beyond their effects upon neuronal physiology and excitability, genetic mutations may also cause plastic changes in the brain, increasing the susceptibility to seizures, as shown by Chiu et al. . In their studies, mice with constitutive expression of the γ2R43Q allele had lower seizure threshold to pentylenetetrazole seizures in adulthood compared with mice conditionally expressing the mutant allele only after the weaning age .
At present, our understanding of genotype–phenotype correlations is very limited to allow definite conclusions. The cost associated with genetic testing and the bias in screening only patients who agree to be tested and have a strong positive family history for certain types of seizures are prohibiting factors in attempts to correctly estimate the prevalence, establish more powerful genotype–phenotype correlations, as well as clarify the interactions with genetic or epigenetic modifiers that may skew such associations . Hopefully, multicenter ongoing studies, as well as the advances from the experimental models of genetic epilepsies, whether conditional or constitutive, will greatly enhance our ability to better understand the underlying pathogenetic mechanisms.
How can we translate these findings to improve the clinical management of patients with genetic epilepsies? Knowing the pathogenic mutation and its functional consequences may potentially help us select or design individualized therapies. For instance, the γ2K289M mutant allele may result in increased potentiation by barbiturates, suggesting that perhaps the utilization of barbiturates in carriers of this mutation may be beneficial, if their seizure types are appropriate for this drug. The availability of drugs that read through premature termination codons may be of use in patients with such mutations. However, it will be essential to determine what will the function of the novel transcript be and what will its impact be on the overall cell physiology. As prior studies indicate , the answer will be different for each mutation. It will therefore be essential to screen the utility and benefit of such drugs in in vitro and in vivo models of genetic diseases before committing them to clinical trials.
Several mutations of the GABAA receptor subunits, the CLCN2 and KCC3 genes have been found in patients with epilepsies of seizures. At present, the pathogenic role of CLCN2 mutations is still unclear, and at best they might contribute to increased susceptibility to seizures, in the presence of additional risk factors. More solid evidence exists on the role of GABAA receptor mutations and KCC3, most of which produce a loss-of-function state due to the production of nonfunctional or hypofunctional proteins, or impair trafficking and insertion to the plasma membrane. However, these affect a minority of the population affected with the associated seizure types and epilepsy syndromes. A better understanding of genotype–phenotype of the affected patients, as well as of the pathogenetic mechanisms leading to ictogenesis and epileptogenesis in these predisposed patients will be invaluable both to improve the management and counseling of these patients and their families, as well as enhance our ability to effectively treat them. The availability of animal models with knockin mutations will be invaluable in such a task.
I would like to acknowledge the funding by NIH NINDS research grants NS20253, NS58303, and NS45243, NINDS/NICHD grant NS62947, as well as grants from People Against Childhood Epilepsy, the International Rett Syndrome Foundation, Johnson & Johnson (unrelated to this paper), and the Heffer Family Foundation that supported research in my lab.