OUP user menu

Dysfunction of the brain calcium channel CaV2.1 in absence epilepsy and episodic ataxia

Paola Imbrici, Stephen L. Jaffe, Louise H. Eunson, Nicholas P. Davies, Colin Herd, Robert Robertson, Dimitri M. Kullmann, Michael G. Hanna
DOI: http://dx.doi.org/10.1093/brain/awh301 2682-2692 First published online: 13 October 2004

Summary

The molecular basis of idiopathic generalized epilepsy remains poorly understood. Absence epilepsy with 3 Hz spike–wave EEG is one of the most common human epilepsies, and is associated with significant morbidity. Several spontaneously occurring genetic mouse models of absence epilepsy are caused by dysfunction of the P/Q-type voltage-gated calcium channel CaV2.1. Such mice exhibit a primary generalized spike–wave EEG, with frequencies in the range of 5–7 Hz, often associated with ataxia, evidence of cerebellar degeneration and abnormal posturing. Previously, we identified a single case of severe primary generalized epilepsy with ataxia associated with CaV2.1 dysfunction, suggesting a possible link between this channel and human absence epilepsy. We now report a family in which absence epilepsy segregates in an autosomal dominant fashion through three generations. Five members exhibit a combination of absence epilepsy (with 3 Hz spike–wave) and cerebellar ataxia. In patients with the absence epilepsy/ataxia phenotype, genetic marker analysis was consistent with linkage to the CACNA1A gene on chromosome 19, which encodes the main pore-forming α1A subunit of CaV2.1 channels (CaV2.1α1). DNA sequence analysis identified a novel point mutation resulting in a radical amino acid substitution (E147K) in CaV2.1α1, which segregated with the epilepsy/ataxia phenotype. Functional expression studies using human CACNA1A cDNA demonstrated that the E147K mutation results in impairment of calcium channel function. Impaired function of the brain calcium channel CaV2.1 may have a central role in the pathogenesis of certain cases of primary generalized epilepsy, particularly when associated with ataxia, which may be wrongly ascribed to anticonvulsant medication.

  • calcium channel
  • epilepsy
  • channelopathy
  • ataxia
  • AEA = absence epilepsy with ataxia
  • AED = anti-epileptic drug
  • EA2 = episodic ataxia type 2
  • EGFP = enhanced green fluorescent protein

Introduction

The neuronal channelopathies are an expanding group of single gene disorders (Kullmann et al., 2002). Among voltage-gated ion channels, mutations of brain potassium (Charlier et al., 1998; Singh et al., 1998) sodium (Wallace et al., 1998; Escayg et al., 2000a; Sugawara et al., 2001) and chloride channels (Haug et al., 2003) have been identified both in relatively uncommon forms of epilepsy and also in a few families with apparently typical idiopathic generalized epilepsy. Childhood absence epilepsy has also been linked to mutations of the T-type calcium channel gene CACNA1H (Chen et al., 2003; although see Heron et al., 2004). The full impact of ion channel dysfunction in the idiopathic generalized epilepsies, including absence epilepsy, is, however, incompletely understood.

Several spontaneously occurring homozygous mouse mutants represent good models for human absence epilepsy, notably tottering, leaner, rocker, rolling Nagoya, lethargic, ducky and stargazer (Burgess and Noebels, 1999; Fletcher et al., 1999; Felix et al., 2002). These mice harbour mutations in the genes coding for subunits making up brain CaV2.1 (also known as P/Q-type) voltage-gated Ca2+ ion channels (Mori et al., 1991; Catterall et al., 2000), which are strongly expressed in the cell bodies and dendrites of cerebellar Purkinje and granule cells (Westenbroek et al., 1995), but also mediate much of the calcium influx that triggers transmitter release at synapses throughout the nervous system. Several of these strains exhibit episodes of motor arrest with spike–wave EEG similar to that seen in human absence epilepsy, but in addition have cerebellar degeneration, ataxia and dystonia. Interestingly, these models also share some features with another autosomal dominant disease, episodic ataxia type 2 (EA2). EA2 is characterized by severe and prolonged attacks of cerebellar ataxia and dysarthria often associated with diplopia or headache. Patients often develop a progressive interictal cerebellar syndrome due to cerebellar damage, which is supported by MRI evidence of vermian atrophy (Griggs et al., 1978; Vighetto et al., 1988). EA2 is caused by several different mutations in CACNA1A, localized on chromosome 19p, which codes for CaV2.1α1, the α1A subunit of the CaV2.1 Ca2+ channel (Ophoff et al., 1996; Denier et al., 1999; Yue et al., 1997, 1998).

We previously have suggested a link between this calcium channel and human absence epilepsy occurring in combination with cerebellar ataxia: we described a singleton case of a child harbouring a novel heterozygous CACNA1A mutation, who showed a phenotype characterized by the early onset of absence epilepsy and cerebellar ataxia (Jouvenceau et al., 2001). The mutation gave rise to a premature stop codon, leading to the smallest truncation yet observed, predicted to lead to the loss of the cytoplasmic C-terminus of the CaV2.1α1 subunit. Heterologous expression of the mutated CaV2.1α1 subunit indicated a complete loss of function, and therefore a profound decrease in Ca2+ current. Because this mutation was a de novo event in a single patient, this case left unanswered the question of whether CACNA1A mutations play a significant role in human absence epilepsy. A further mutation in CACNA1A, predicted to result in a frameshift and premature stop codon, has been reported in association with EA2 and epilepsy, but without details of the seizure type (Jen et al., 2004).

We now describe a new family in which six members in three generations exhibited a typical primary generalized 3 Hz spike wave EEG abnormality. Five of these individuals exhibited clinical absence epilepsy with ataxia (AEA). All the AEA cases harboured a novel point mutation in the CACNA1A gene, which impaired calcium channel function.

Methods

Linkage analysis

Genomic DNA was obtained from 11 family members, including five with the AEA phenotype and one with EEG changes only. DNA was extracted from white blood cells using standard methodology (Miller et al., 1988). Individuals were genotyped for seven polymorphic microsatellite markers using fluorescent-based semi-automated methodology on an ABI 373A (Applied Biosystems). The microsatellites span 1.9 Mb encompassing CACNA1A, and include three CACNA1A intragenic polymorphisms: AAG18 (Trettel et al., 2000) (intron 1), D19S1150 (Ophoff et al., 1996) (intron 7) and A1ACAG (Trettel et al., 2000) (CAG repeat in the 3′-untranslated region). Allele sizes were determined using Genotyper software (Applied Biosystems). The power of the family resource to detect linkage was calculated using SLINK (Ott, 1989; Weeks et al., 1990) under the assumption of autosomal dominant inheritance with 100% penetrance. The multipoint LOD score was calculated using GeneHunter (Kruglyak et al., 1996).

Mutation detection

DNA was extracted from all available family members using standard procedures. Each exon and the flanking intronic regions of CACNA1A was amplified by polymerase chain reaction (PCR). Each exon was then analysed by direct DNA sequencing on an ABI 377 Sequencer (Applied Biosystems, Foster City, CA). Primers and PCR conditions are available from the authors on request. Restriction fragment length polymorphism analysis was used to screen 200 control chromosomes and to confirm segregation of the mutation with disease in the family.

cDNA construction

The E147K mutation identified in this study (see Results) was introduced into human CACNA1A cDNA (CaV2.1α1) with a PCR-based strategy using proofreading DNA polymerase (Pfu, Promega). The cDNA was the same as used by Hans et al. (1999) and Guida et al. (2001). Four primers were used in three separate PCRs. Forward primer 1 (f1) TGTCAAGCTTGCGGTGTGGCAGG, forward primer 2 (f2) GGAATTTTTTGTTTCAAGGCTGGAATTAAAATC, reverse primer 1 (r1) GATTTTAATTCCAGCCTTGAAACAAAAAATTCC and reverse primer 2 (r2) TAGCAACACACAGCGGTGTTGAG. Primers f1 and r2 flank the region of interest and lie 5′ to a BglII restriction site and 3′ to a DraIII restriction site, respectively. Primer pairs f1/r1 and f2/r2 were used in two separate reactions on hα1A in expression plasmid pMT2 LF. The products from these two reactions were used in a third reaction along with primers f1 and r2 to yield the full-length product. This was digested with the enzymes BglII and DraIII and subcloned into CACNA1A-pMT2 LF, which had been cut accordingly. The sequence of the mutated CACNA1A cDNA was verified using a fluorescent BigDye terminator sequencing kit (Applied Biosystems).

For the construction of the enhanced green fluorescent protein (EGFP)–CaV2.1α1 fusion plasmid, the entire coding sequence of either the wild-type or the mutant channel (BamHI–SalI) was subcloned into the BglII–SalI sites of the pEGFP-C3 expression vector (Clontech). This generates a fusion protein with the EGFP sequence appended to the N-terminus of the channel.

Functional expression in oocytes

Female Xenopus laevis frogs were killed with 0.5% tricaine, decapitation and pithing. Stage V–VI oocytes were isolated and stored at 4–18°C in fresh ND96 medium (NaCl 96 mmol/l, KCl 2 mmol/l, MgCl2 1 mmol/l, CaCl2 1.8 mmol/l, HEPES 5 mmol/l, gentamicin 50 µg/ml pH 7.4; sterilized by filtration).

CACNA1A cDNA was injected into the nuclei of the oocytes (Nanoject, Drummond, Broomall, PA) together with cDNA encoding the auxiliary subunits β4 and α2δ1 (provided by A. Dolphin, University College London). cDNA (9.2 nl) was injected at a concentration of 0.6–1.0 mg/ml. For wild-type or mutant expression, cDNAs encoding CaV2.1α1, β4 and α2δ1 were injected in a ratio of1 : 1 : 1. For co-expression experiments, wild-type CaV2.1α1, mutant CaV2.1α1, β4 and α2δ1 were injected in a ratio of 1 : 1 : 2 : 2. The total concentration of cDNA, determined from optical density, was constant in each case and the results were normalized by the wild-type amplitude recorded on the same day within the same batch of oocytes. Normalized amplitudes were then averaged between batches.

Whole-cell currents were measured with a two-electrode voltage clamp 3–4 days after injection (Geneclamp 500B, Axon Instruments, Union City, CA). Currents were filtered at 1 kHz and sampled at 5 kHz. The perfusion solution used to record Ba2+ currents contained Ba(OH)2 (40 mmol/l), NaOH (50 mmol/l), KOH (1 mmol/l), niflumic acid (0.4 mmol/l) and HEPES (10 mmol/l), with pH adjusted to 7.4 with methanesulphonic acid, and was kept at room temperature. The voltage dependence of activation was obtained by plotting the normalized tail currents recorded at −50 mV as a function of the pre-pulse potential (from −80 to +65 mV). The voltage dependence of inactivation (steady-state inactivation) was determined from the normalized inward current elicited during steps to +10 mV after 10 s steps to various holding potentials (from −140 to +40 mV). Data were fitted with the Boltzmann function: I = 1/[1 + exp{−(V − V1/2)/k}] from which V1/2 and the slope factor k, for activation and steady-state inactivation, were computed. Inactivation during depolarization was estimated during 3 s pulses from a holding potential of −90 mV to a test potential of +10 mV. Traces were fitted with a double exponential function from which the time constants and relative amplitudes of the fast (τfast, %fast) and slow (τslow, %slow) components were obtained. The rate of current activation was estimated from the 10–90% rise time for currents evoked by test pulses ranging from −80 to +40 mV from a holding potential of −90 mV.

Data were acquired and analysed with Pclamp6 (Axon Instruments) and Origin 6.0 (Microcal). Significance was tested with unpaired t tests with Bonferroni correction for multiple comparisons.

Confocal laser scanning microscopy

Oocytes were injected with cDNA encoding EGFP–CaV2.1α1, β4 and α2δ1 as above, with the same ratios of wild-type and mutant CACNA1A. Following incubation for 3 days, they were fixed in 4% paraformaldehyde [v/v in 0.15 M phosphate-buffered saline (PBS)] for at least 3 h at 4°C, and were then cut in half, perpendicular to the equator. Each half was submerged in ND96 medium and imaged on the cytoplasmic side, followed by the membrane side, obtaining four images per oocyte.

Fluorescence images were acquired at 10× magnification using a Bio-Rad Radiance2000 laser scanning confocal microscope with an 8 bit ADC, using Lasersharp software (Bio-Rad) for acquisition and Scion Image software (NIH) for analysis. EGFP was excited using a 488 nm argon laser beam, with an emission filter of 515/30 nm to collect the peak emission (507 nm). Laser and confocal aperture settings were kept constant for all experiments. A Z-stack was acquired at 10 µm steps and then projected onto a single image for analysis.

The images taken from the cytoplasmic aspect were subdivided into (i) an annulus corresponding to the membrane and immediately submembrane area (7 µm from the edge) and (ii) the remainder of the cytoplasm, including nucleus and juxta-nuclear endosomes. The mean pixel value within each region was divided by its area, and then averaged with the measurement obtained from the second half of the oocyte, to give a mean fluorescence value. The membrane fluorescence value obtained from the surface view was added to that obtained in the membrane annulus of the cytoplasmic view, to obtain an estimate for the fluorescence associated with the plasmalemma. The cytoplasmic mean fluorescence values for each cell were then averaged between oocytes. As a measure of total cell fluorescence, the membrane and cytoplamsic data were first averaged for each cell, and then among all cells.

Results

Clinical data

We ascertained a large North American pedigree in which five members exhibited absence epilepsy (3 Hz spike–wave EEG) with variable degrees of episodic and anticonvulsant drug-induced cerebellar ataxia; we have used the abbreviation ‘AEA’ for this phenotype. One individual had a typical 3 Hz spike–wave EEG but had no clinical phenotype, and was neurologically normal (III:2 below). This family has been assessed in detail and followed-up since 1988 by one of us (S.L.J.). We performed clinical, EEG and genetic analyses on 11 genetically related individuals and one non-blood relative who had married into the family. Brain imaging was also undertaken in certain cases. The key clinical and investigative details of these individuals are shown in Tables 1 and 2.

View this table:
Table 1

Clinical details of affected family members

Pedigree identifierAge of onsetAbsence epilepsyEpisodic ataxiaOther featuresEEG
I:2TeensYesYesMild CA3 Hz s/w
II:557 yearsNoNoMild CA3 Hz s/w
II:613 yearsYesYesNo CA3 Hz s/w
II:710 yearsNoYesCAA3 Hz s/w
III:34 yearsYesYesCA and CAA3 Hz s/w
  • CA = clinical signs of cerebellar ataxia; CAA = significant worsening of cerebellar ataxia on exposure to therapeutic doses of AEDs; s/w = spike–wave.

View this table:
Table 2

Clinical assessment of unaffected family members

Pedigree identifierAge at exam (years)Absence epilepsyEpisodic ataxiaOther featuresEEG
II:155NoNoNoneNormal
II:256NoNoNoneNormal
II:358NoNoNoneNormal
II:459NoNoNoneNormal
II:1062NoNoNoneNormal
III:127NoNoNoneNormal
III:220NoNoNone3 Hz s/w
  • Individual II:10 is a non-blood relative marrying into the pedigree. s/w = spike–wave.

The range of the clinically penetrant phenotype is illustrated by the following two case histories.

Case II:6 (index case) presented at the age of 39 years with a history of ‘dizzy’ attacks stretching back to teenage years. This had worsened in the year prior to assessment. A typical attack consisted of sudden onset disequilibrium with associated headache and nausea. The patient was unable to walk during the attacks because of ataxia and had to lie down. Attacks lasted between 1 and 8 h, and could be precipitated by stress and anxiety.

In addition, there was a history of blank episodes suggestive of absence seizures from late childhood which subsided in adult life when the ataxic episodes were the major clinical problem. In the absence attacks, typically there would be motor arrest without warning, lasting seconds. Postural tone was not lost. Generally, the patient was able to resume normal activities immediately. On other occasions, there were more prolonged atypical absence episodes. During one EEG examination, the patient developed a prolonged fugue-like state lasting several hours accompanied by generalized 3 Hz spike–wave discharges. A CT brain scan revealed mild midline cerebellar atrophy, but neurological examination was normal. An MRI of the brain confirmed the presence of cerebellar vermian atrophy. Her symptoms remitted on a combination of acetazolamide and carbamazepine. However, it was particularly notable that she exhibited marked cerebellar ataxia when antiepileptic drug (AED) levels were even in the mid-therapeutic range. She was given a diagnosis of EA2 with absence epilepsy.

Case III:3 exhibits a more severe younger onset phenotype with striking similarities to the original case of EA2/epilepsy reported by Jouvenceau et al. (2001). Absence attacks characterized by motor arrest without warning developed from age 4 years. Attacks were very brief (seconds) and apparently normal behaviour resumed. No clonic jerking or automatisms were reported during the episodes. Early motor milestones were considered delayed, being a ‘clumsy walker’ and not ambulating until age 2.5 years. An initial EEG confirmed a polyspike–wave primary generalized abnormality; further EEGs demonstrated generalized 3 Hz spike–wave changes. His seizures were not easily controlled by several AEDs including clonazepam and phenytoin. It had been noted that he was considered very ‘sensitive’ to AEDs with a tendency to develop marked cerebellar ataxia even with low therapeutic doses of phenytoin and other medications. On one occasion, he developed extreme cerebellar ataxia and a decreased level of consciousness. Although there was no clear-cut history of episodes of cerebellar ataxia, examination at the age of 10 years revealed moderate fixed cerebellar signs. He had some learning disabilities and was considered to have cognitive impairment, but no clear speech difficulty. He was considered to require special education which he entered at the age of 8 years. It is considered that he will not achieve independent employment. Unfortunately, formal neuropsychometry is not available.

Genetic studies

Linkage analysis

The maximum simulated LOD score was 2.4 at θ = 0. Five of the seven microsatellite markers were informative in this family. The five affected individuals shared a single common haplotype between D19S914 and D19S840 (Fig. 1). This haplotype was not shared by any of the unaffected individuals. The maximum multipoint LOD score across this haplotype was 2.1. The AEA phenotype in this pedigree was thus consistent with linkage to CACNA1A.

Fig. 1

Haplotypes of AEA family members using seven microsatellite markers spanning 1.9 Mb encompassing CACNA1A. The ‘disease’ haplotype is outlined. The pedigree identifiers are listed.

CACNA1A gene DNA sequence analysis

We identified a heterozygous point mutation (G439A) in exon 3 of CACNA1A, which results in a substitution of lysine for glutamic acid at codon 147 (E147K). This previously unreported change segregated with the AEA disease phenotype. The mutation was not detected in unaffected individuals or in the patient with an abnormal EEG but no clinical phenotype [case III:2]. All family members were sequenced. The mutation identified results in the loss of a TaqI restriction site. With wild-type exon 3, the 203 bp product is cut into a 165 bp fragment and a 38 bp fragment (the latter runs off the gel). With the heterozygous E147K mutation, an uncut fragment at 203 bp is seen in addition to the band at 165 bp (Fig. 2). This mutation was absent from a panel of 200 control chromosomes, and segregation of the mutant with the AEA disease was confirmed in the family (Fig. 1). The mutation affects a highly conserved residue within the second transmembrane segment of domain I of CaV2.1α1, as assessed by an interspecies comparison (Fig. 3).

Fig. 2

Upper panel: family pedigree of the absence epilepsy EA2 family. Affected individuals are shown as filled symbols and unaffected individuals as unfilled symbols. Lower panel: results of restriction fragment length polymorphism analysis for all individuals shown in the pedigree. The presence of the E147K mutation results in two bands on the gel as described in the text. All affected individuals exhibit two bands; all unaffecteds exhibit one band. C = wild-type control DNA sample. Electropherogram of DNA sequences indicating the G to A transition resulting in the E147K amino acid substitution.

Fig. 3

Conservation of E147 across species and other human calcium α1 subunits.

Functional consequences of the E147K mutation

Electrophysiology

To investigate the functional consequences of the E147K mutation, wild-type and/or mutant CaV2.1α1 were expressed in Xenopus oocytes, together with auxiliary subunits α2δ1 and β4. Wild-type CaV2.1α1 expression gave typical CaV2.1 currents elicited with a depolarizing ramp protocol from −90 to +90 mV. The mutant CaV2.1α1 was functional but showed reduced current expression (P < 0.0005; Table 3 and Fig. 4A). Jouvenceau et al. (2001) previously reported that the R1820X mutation, associated with epilepsy, has a dominant-negative effect on wild-type CaV2.1α1, when inserted into the rabbit BI-1 cDNA encoding CaV2.1α1 (although this effect was not seen when inserted into the human cDNA; P. Imbrici and D. M. Kullmann, unpublished observations). We therefore co-expressed E147K with wild-type CaV2.1α1 (together with the accessory subunits). The peak current amplitude mediated by co-expressed wild-type and mutant CaV2.1α1 was consistent with linear summation of the currents mediated by each subunit separately (Fig. 4B; Table 3). These results thus argue that the mutation causes a decrease in current amplitude but without evidence for a dominant-negative interaction with the wild-type subunit.

View this table:
Table 3

Kinetic parameters

Wild-typeE147KWT + E147K
Peak current (μA)2.8 ± 0.5, n = 451.0 ± 0.2, n = 44**3.3 ± 0.4, n = 50
Apparent reversal potential (mV)49.5 ± 0.8, n = 3949.4 ± 1.0, n = 4348.3 ± 1.0, n = 29
Voltage dependence of activation
    Vh (mV)−13.7 ± 0.9−9.0 ± 1.0−14.9 ± 1.0
    k (mV)5.7 ± 0.5, n = 255.0 ± 0.7, n = 195.1 ± 0.4, n = 21
Voltage dependence of inactivation
    Vh (mV)−32.9 ± 0.7−35.4 ± 0.7−35.1 ± 0.7
    k (mV)15.8 ± 0.3, n = 914.0 ± 0.4, n = 1014.3 ± 0.4, n = 11
Inactivation rate (10 mV)
    τfast (ms)245 ± 23 (24%)280 ± 40 (27%)310 ± 12 (27%)
    τslow (ms)1080 ± 147 (76%), n = 241817 ± 241 (73%), n = 15*1372 ± 133 (73%), n = 12
Weighted τ (ms)960 ± 591459 ± 199*1099 ± 141
Activation rate (5 mV)
    10–90% rise time (ms)5.1 ± 0.5, n = 106.9 ± 1.8, n = 155.2 ± 1.1, n = 12
  • * P < 0.05

  • ** P < 0.0005 relative to wild-type.

Fig. 4

Results of Xenopus oocyte expression studies. (A) Representative current traces from oocytes injected with wild-type and mutant E147K, together with auxiliary subunits, in response to a depolarizing ramp from −90 to +90 mV. (B) Amplitude histogram showing the whole-cell currents evoked at voltages giving maximal currents. The current amplitudes were normalized to wild-type (WT1X). The bar labelled WT2X indicates the current amplitude obtained by doubling the amount of wild-type cDNA injected. This shows that the current density was approximately linearly related to the availability of cDNA encoding CaV2.1α1. (C) Amplitude histograms of whole-cell currents from oocytes injected with wild-type and mutant together with a double amount of accessory subunits. Data are mean ± SEM for four different batches of oocytes (*significantly different from wild-type, P < 0.05).

The accessory subunits α2δ1 and β4 contribute to targeting of calcium channels to the plasma membrane, and also affect their biophysical properties (de Waard et al., 1995; Walker et al., 1998). We looked for evidence that the mutation interferes with these interactions by increasing the amount of cDNA encoding the accessory subunits. This had no effect on the wild-type current amplitude, arguing that the amount of CaV2.1α1 limits the number of channels reaching the membrane. In contrast, doubling the amount of α2δ1 and β4 cDNA led to an increase in current mediated by channels containing the mutant CaV2.1α1 (Fig. 4C). Indeed, the current reached the amplitude of wild-type.

The effect of increasing the availability of α2δ1 and β4 is consistent with impaired membrane targeting, which can be overcome by increasing the abundance of accessory subunits. However, because accessory subunits also affect calcium channel kinetics, and because impaired activation of CaV2.1 could also explain the results, we examined the effect of the E147K mutation on the voltage- and time-dependent kinetics of the channel (Figs 5 and 6; Table 3). The voltage dependence of activation for the mutant channel, measured with depolarizing membrane potential steps, showed a small shift towards more positive potentials, although this failed to reach significance (Fig. 5A and B). When co-expressed with the wild-type subunit, this effect on activation threshold disappeared, consistent with the fact that most of the current was mediated by wild-type channels. There was no significant difference in the apparent reversal potential (Table 3). The voltage dependence of steady-state inactivation also did not differ among oocytes expressing wild-type channels, mutant channels or both (Fig. 5C).

Fig. 5

(A) Sample traces obtained for wild-type alone, wild-type co-expressed with E147K and E147K alone in response to depolarizing steps (−60 to +30 mV). (B) Voltage dependence of activation for wild-type alone (▪), wild-type co-expressed with mutant (○) and mutant E147K alone (•). The mutant showed a non-significant shift to depolarized voltages. (C) Voltage dependence of steady-state inactivation.

Fig. 6

(A) The 10–90% rise time plotted against test potentials for wild-type alone, wild-type co-expressed with mutant and mutant E147K alone (symbols as in Fig. 5). (B) Sample current traces recorded from −90 to +10 mV for 3 ms, used to estimate inactivation kinetics. (C) Inactivation time constants (τfast and τslow) for wild-type alone, wild-type co-expressed with mutant and E147K alone. The E147K mutation was associated with a significant increase in τslow (P < 0.05).

The rate of activation, estimated from the 10–90% rise time, showed a non-significant slowing for the mutant (Fig. 6A). Finally, we examined the rate of activation and inactivation evoked by 3 s pulses to different voltages between −20 and +30 mV (Fig. 6B and C). Inactivation was fitted by a double exponential time course. The slow component of inactivation showed a small prolongation relative to wild-type, which reached significance at P < 0.05 (Table 3). This was also seen for the weighted time constant (calculated as τfast %fast + τslow %slow).

Membrane expression of EGFP–CaV2.1α1

The functional analysis of the E147K mutation indicates a partial loss of function, which can be rescued by increasing the availability of accessory subunits, and a small slowing of inactivation. Because the most robust effect could be explained by impaired membrane targeting, we examined the fluorescence intensity of oocytes expressing mutant or wild-type EGFP–CaV2.1α.1 or uninjected oocytes.

We first measured the biophysical properties of the fusion proteins, to determine whether the presence of EGFP at the N-terminus has an effect on current density, or voltage- and time-dependent kinetics. The EGFP tag did not interfere with the function of the channel. There were no differences in current density, current–voltage relationship and rate of activation. We then assessed the fluorescence distribution in the oocytes (Fig. 7). Oocytes show considerable autofluorescence, which was measured in uninjected oocytes to permit analysis of the effects of expressing EGFP–CaV2.1α1. The membrane mean fluorescence of oocytes injected with mutant EGFP–CaV2.1α1 was significantly reduced compared with wild-type EGFP–CaV2.1α.1-injected oocytes (Fig. 8A). However, the cytoplasm mean fluorescence measured for the same cells was not different (Fig. 8B). These data suggest that a reduction in the number of mutant channels reaching the membrane could account for the reduced current level.

Fig. 7

Example images acquired by fluorescence imaging of the cytoplasmic and membrane sides of halved oocytes injected with EGFP–CaV2.1α.1 (top row) or EGFP–E147K (middle row) and uninjected oocytes (bottom row). Dotted lines indicate the regions of fluorescence measurement: for the cytoplasmic face, the annulus corresponding to the edge of the oocyte was considered as membrane fluorescence.

Fig. 8

Bar charts showing the mean fluorescence pixel value (F, arbitrary units) measured for the membrane (A) and for the cytoplasm (B), obtained for oocytes imaged on the membrane and cytoplasmic sides. Oocytes injected with EGFP–E147K (n = 26) and uninjected oocytes (n = 16) showed a significantly (P < 0.05) reduced membrane fluorescence compared with EGFP–CaV2.1α1 (n = 24). Membrane fluorescence (arbitrary units – Wan ± SEM): WT, 1385 ± 53. E147K, 1186 ± 52; uninjected, 1089 ± 46. Cytoplasmic fluoresence: WT, 750 ± 29; E147K, 714 ± 42; uninjected, 707 ± 46.

Discussion

This is the first reported family in which typical absence seizures co-segregate with a CACNA1A mutation that impairs Cav2.1 function. These data strongly argue that dysfunction of Cav2.1, observed in several mouse models of absence epilepsy, plays a role in the aetiology of human absence epilepsy.

Affected members in this family exhibited a wide range of phenotypic severity, despite the presence of the same pathogenic mutation. In some cases, the ataxia component of the phenotype was indistinguishable from EA2, whereas in others the major feature was marked ataxia in response to AEDs at normally non-toxic doses. In all affected cases, the EEG showed classical 3 Hz spike–wave discharges characteristic of typical absence epilepsy. Although the phenotypic severity was variable in this family, it was generally milder than in a singleton case with a de novo CACNA1A mutation that first suggested that Cav2.1 dysfunction can underlie human absence epilepsy (Jouvenceau et al., 2001). This case had severe very progressive ataxia, poorly controlled epilepsy and mental retardation. In contrast, affected members in the present family have a generally milder phenotype often indistinguishable from commonly encountered absence epilepsy, which affects ∼0.5% of all children. Generally, the absence epilepsy in this family was controlled with medication. It is notable that when ataxia is encountered in such children, it is commonly ascribed to AED toxicity.

The missense mutation identified in the present family gives rise to an amino acid substitution (E147K) in the second transmembrane segment of domain I of CaV2.1α1. Although this is a highly conserved residue, it had a relatively subtle effect on the function of the channel, compared with the premature stop codon (R1820X) identified in the previously reported case. R1820X led to a complete loss of function and, when inserted in the rabbit cDNA BI-1, it also showed a dominant-negative effect on the wild-type subunit, similar to that reported for experimental truncations of the closely related subunit CaV2.2α1 (Raghib et al., 2001). The molecular mechanism of this interaction is unclear, but may involve misfolding of the wild-type protein when it encounters the mutant peptide. Moreover, although we subsequently have confirmed that the R1820X mutation is non-functional when inserted in the human CaV2.1α1 cDNA, the dominant-negative effect is no longer seen, suggesting that it may depend on the splice variant represented by the cDNA chosen for expression (P. Imbrici and D. M. Kullmann, unpublished). In contrast to these severe effects on channel function, the mutation identified in the present family only produced a partial reduction in calcium channel function, which could be rescued by overexpression of accessory subunits. The results of EGFP fluorescence imaging imply that the mutation impairs trafficking to the membrane. We cannot, however, exclude an effect on channel opening probability, although the kinetic measurements only revealed a modest slowing of inactivation.

We tentatively suggest that the main deficit in the affected members of this family is a reduction in membrane expression of CaV2.1. This is subject to the caveat that it is not known whether availability of accessory subunits is a limiting factor in membrane expression in vivo. Moreover, CACNA1A undergoes extensive splicing (which may underlie the discrepancies in R1820X–wild-type interaction that we have observed previously), further confounding the interpretation of the results. If the major effect is a decrease in membrane expression, the milder effect in vitro accords with the generally milder disease pattern than seen for R1820X. Impaired trafficking of the mutant CaV2.1α1 contrasts with the effects of other mutations associated with episodic ataxia without epilepsy. These have been reported either to be non-functional (with, in one case, intact membrane expression; Guida et al., 2001) or to impair channel function, either by shifting the voltage dependence of activation to more positive potentials or by reducing the open channel probability (Wappl et al., 2002).

A decrease in CaV2.1 current density is also seen in cerebellar neurons studied in acute brain slices from several homozygous mice with mutations of genes encoding CaV2.1α1, β4 and α2δ2 (Wakamori et al., 1998; Jun et al., 1999; Lin et al., 1999; Barclay et al., 2001; Fletcher et al., 2001). These mice frequently have ataxia and behavioural arrest with spike–wave EEG, reminiscent of the AEA phenotype of the present family. Several of these strains show cerebellar degeneration, also seen in the patients, which presumably does not result from an excessive Ca2+ influx via CaV2.1 channels but instead may arise from a decrease. However, compensatory alterations in other high threshold calcium channels have also been reported in mutant mice (Campbell et al., 1999; Fletcher et al., 2001). A toxic effect of mutations on cerebellar neurons is unlikely to explain either the paroxysmal nature of the ataxiadoes or the occurrence of absence seizures, which are thought to originate from imbalance of the thalamocortical circuitry (Huntsman et al., 1999). Some preliminary work from mutant mice implies that CaV2.1 mutations disturb the balance of GABA and glutamate release from presynaptic boutons (Caddick et al., 1999). Spontaneous mutations in CaV2.1 subunits associated with spike–wave seizures also show an increase in low threshold Ca2+ currents mediated by T-type channels (Zhang et al., 2002). If this also occurs in association with the E147K mutation, it provides a compelling mechanism for the initiation of absence-type seizures in the present family: human mutations of CACNA1H, encoding the T-type channel subunit CaV3.2α1, have been identified in sporadic cases of childhood absence epilepsy (Chen et al., 2003), and some of these have been shown to lead to a gain of function of this channel subtype (Khosravani et al., 2004). T-type channels are also a target of ethosuximide (Gomora et al., 2001), which, although not tried in the family described here, has a useful anti-absence profile.

The observation of a single member of the present family with a classical 3 Hz spike–wave EEG but with no epilepsy or ataxia phenotype requires further consideration. This individual did not have the E147K mutation, arguing for another cause for the EEG abnormality. (The absence of the mutation was confirmed on two further occasions from two separate blood samples, ruling out sample mislabelling.) One important possibility is that a second gene variant was inherited in this family, which, on its own, gives rise to the EEG abnormality, but is insufficient to give rise to absence seizures, let alone episodic ataxia. We cannot exclude the possibility that this variant was also inherited by the affected members, and that the AEA phenotype is actually a manifestation of digenic inheritance, as has been suggested to occur in another family with epilepsy (Baulac et al., 2001). If a second gene was indeed inherited in this family, what is the probability that it, on its own, was responsible for the abnormal EEG, and not the E147K mutation? A simple calculation argues strongly that this is very low: the probability that the allele corresponding to the abnormal EEG would be inherited from case I:1 by the affected cases in the second and third generations, and not by the unaffected cases (excluding the unaffected branch of the family III:1) is 1 : 29 or <0.002.

In conclusion, this family provides strong evidence that dysfunction of Cav2.1 is implicated in certain cases of human absence epilepsy, and adds to previous indirect suggestions for such a link from mutant mice, from previous sporadic cases (Escayg et al., 2000b; Jouvenceau et al., 2001) and from linkage disequilibrium associating idiopathic generalized epilepsy with the CACNA1A locus (Chioza et al., 2001). From a clinical viewpoint, we suggest that the presence of cerebellar ataxia, either episodic or in response to moderate doses of AEDs, increases the liklehood that CaV2.1 dysfunction may be involved. CaV2.1 variants may act in combination with other as yet unidentified genetic factors to produce the clinical phenotype of AEA in combination with a primary generalized 3 Hz spike wave disturbance.

Acknowledgments

We wish to thank K. A. Stauderman for the gift of the CACNA1A, A. Dolphin for α2δ1 and β4 and the expression vector pMT2 LF, and to S. Maltas for assistance with molecular biology. Supported by the University College Hospitals Special Trustees, Wellcome Trust and MRC.

Footnotes

  • * These authors contributed equally to this work

References

View Abstract