Background Early infantile epileptic encephalopathies usually manifest as severely impaired cognitive and motor development and often result in a devastating permanent global developmental delay and intellectual disability. A large set of genes has been implicated in the aetiology of this heterogeneous group of disorders. Among these, the ion channelopathies play a prominent role. In this study, we investigated the genetic cause of infantile epilepsy in three affected siblings.
Methods and results Homozygosity mapping in DNA samples followed by exome analysis in one of the patients resulted in the identification of a homozygous mutation, p.L1040P, in the CACNA2D2 gene. This gene encodes the auxiliary α2δ2 subunit of high voltage gated calcium channels. The expression of the α2δ2-L1040P mutant instead of α2δ2 wild-type (WT) in Xenopus laevis oocytes was associated with a notable reduction of current density of both N (CaV2.2) and L (CaV1.2) type calcium channels. Western blot and confocal imaging analyses showed that the α2δ2-L1040P mutant was synthesised normally in oocyte but only the α2δ2-WT, and not the α2δ2-L1040P mutant, increased the expression of α1B, the pore forming subunit of CaV2.2, at the plasma membrane. The expression of α2δ2-WT with CaV2.2 increased the surface expression of α1B 2.5–3 fold and accelerated current inactivation, whereas α2δ2-L1040P did not produce any of these effects.
Conclusions L1040P mutation in the CACNA2D2 gene is associated with dysfunction of α2δ2, resulting in reduced current density and slow inactivation in neuronal calcium channels. The prolonged calcium entry during depolarisation and changes in surface density of calcium channels caused by deficient α2δ2 could underlie the epileptic phenotype. This is the first report of an encephalopathy caused by mutation in the auxiliary α2δ subunit of high voltage gated calcium channels in humans, illustrating the importance of this subunit in normal physiology of the human brain.
- Epilepsy and seizures
Statistics from Altmetric.com
Early infantile epileptic encephalopathies (EIEE) manifest in the neonatal or the early infantile period as severely impaired cognitive and motor development due to recurrent clinical seizures or prominent interictal epileptiform discharges. Because of the susceptibility of the developing brain to epilepsy, the result is often a devastating permanent global developmental delay and intellectual disability. With the introduction of genomic technologies, the importance of a large set of genes in the emergence of EIEE has been unravelled, including those involved with synaptogenesis, pruning, neuronal migration and differentiation, neurotransmitter synthesis and release, and the structure and function of membrane receptors and transporters (for review see Tavyev Asher and Scaglia1). Among these, the ion channelopathies play a prominent role, especially the voltage gated sodium channel SCN1A, with its documented 856 disease-causing mutations (HGMD 2012.2, release date 29 June 2012), which were associated with several EIEE including migrating partial seizures of infancy, severe infantile multifocal epilepsy, and Dravet syndrome.2 ,3 We now report a novel calcium channelopathy identified in three infants with EIEE by a molecular and functional study.
Subjects and methods
The subjects were three sibs, two males and one female (3110, 3111, and 10177 in figure 1A) who suffered from epilepsy and global developmental delay. They were the offspring of healthy Arab-Palestinian first degree cousins. An older daughter (3109) was healthy. The pregnancy, delivery, and perinatal course of the three patients were uneventful. Nonetheless at birth there was generalised muscle hypotonia. Seizures were first noted at 20–60 days of age, initially consisting of eye-rolling and facial twitching, but these were later replaced by several types of seizures including atonic, clonic and tonic attacks without focality. The seizures were resistant to multiple medications including vigabatrin, clonazepam, topiramate, clobazam, phenobarbital, and valproic acid. The earliest electroencephalogram (EEG) available for review was obtained at 10 months of age and was consistent with Lennox–Gastaut syndrome. At 7 years of age, the EEG had evolved into a slow background rhythm with multifocal spikes and slow waves (figure 2).
Physical examination of the patients (the oldest was 7 years of age) revealed well nourished children, with no dysmorphic features and normal head circumference. Axial hypotonia was noted while the appendicular tone was normal. Brisk, symmetric reflexes were obtained without other pyramidal signs. Choreiform movements were noted intermittently. The children made no eye contact but the oldest appeared to track and react to noise; none of them could speak or use their hands purposefully. Repeated examinations over 3 years of follow-up revealed no major changes and the patients remained hypotonic, lacked psychomotor development, and suffered from pharmaco-resistant epilepsy. Growth parameters and general health remained satisfactory and there were no clinical or biochemical indications of the involvement of other systems. A thorough investigation included analysis of plasma and cerebrospinal fluid amino acids, urine organic acids, plasma very long chain fatty acids, isoelectric focusing of transferrins, serum biotinidase and carnitine values, muscle pathology, and mitochondrial respiratory chain enzymes, which were all normal. Chromosomal rearrangements were excluded by Affymetrix Genome-Wide Human Single Nucleotide Polymorphism (SNP) Array 6.0 in patient 3111 DNA. Brain MRI of patients 3111 and 3110 performed at age 4 years and 7 years, respectively, showed paucity of white matter and cerebellar atrophy (figure 2).
A search for common homozygous regions in the DNA of the affected sibs and their unaffected sister was performed using the Affymetrix GeneChip Human Mapping 250K SNP Array, as previously described.4
Whole exome sequencing
Whole exome sequencing was performed in DNA from patient 3111 using the SureSelect Human All Exon V.2 Kit (Agilent Technologies, Santa Clara, California, USA) on HiSeq2000 (Illumina, San Diego, California, USA) as 100 bp paired-end runs. Image analysis and base calling were performed with the Genome Analyser Pipeline version 1.5 with default parameters. The reads were aligned with DNAnexus software (Palo Alto, California, USA) using the default parameters with the human genome assembly hg19 (GRCh37) as reference, as previously described.5
Maintenance of Xenopus laevis female frogs, preparation of oocytes, in vitro RNA synthesis, and measurement of currents using two-electrode voltage clamp were carried out as described.6 Specifically, oocytes were injected 3–4 days before measurement with RNA of α1B (d14157) or α1C (X15539) together with β3 (NM_012828) or β2b (X64297), respectively, along with RNA of α2δ2 (NM_006030)-WT or of α2δ2-L1040P with the single nucleotide mutation t3119c produced by PCR (Roche). Whole cell currents were recorded in Xenopus oocytes using the two-electrode voltage clamp technique, as described.6 A quantity of 30 nl of 50 mM BAPTA (Ca2+ chelator, to avoid endogenous Ca2+ dependent Cl− currents) was routinely injected into the oocytes 0.5–2 h before the measurement of currents. The extracellular solution concentrations were adjusted in each experiment as indicated, to achieve currents ranging from 300–5000 nA using 40, 10 or 2 mM Ba(OH)2 or Ca(OH)2. The extracellular solution contained 50–90 mM NaOH, 2 mM KOH, and 5 mM HEPES, titrated to pH 7.5 with methanesulfonic acid (Sigma). In each oocyte, the net current was obtained by subtraction of the residual currents recorded with the same protocols after applying 200 μm CdCl2 to the same solution. Recording was performed using a GeneClamp 500 amplifier (Molecular Devices). Stimulation, data acquisition and analysis were performed using pCLAMP 10.2 software (Molecular Devices). r400 was calculated as the fraction of current remaining after 400 ms depolarising pulse normalised to the peak current.
Three to four days after injection of RNA, 10–20 oocytes were homogenised on ice in buffer (20 mM Tris, pH 7.4, 5 mM EGTA, 5 mM EDTA, and 100 mM NaCl) containing protease inhibitor (Roche). Debris was removed by 1000×g centrifugation for 15 min at 4°C. Protein samples were separated on SDS-8% polyacrylamide gels. Antibody against α2δ2 (Alomone Labs) was used to detect the protein. Visualisation of protein bands was performed using ECL reagents (Pierce).
Oocytes expressing yellow fluorescent protein (YFP) labelled α1B, YFP-α1B,7 were placed in ND96 solution in a glass bottom dish. Fluorescent signals were collected from the animal hemisphere of the oocyte with a confocal microscope (Zeiss 510 META, 20× lens, digital zoom 2). YFP was excited using a 514 nm laser. Emission was collected in the 529–538 nm range in the Meta mode, and used for comparison of expression levels.
Results are shown as mean±SE; numbers within the bars represent n, number of cells measured in the same group. Two group comparisons were done using Student's t test. Asterisks indicate statistically significant differences as follows: *p<0.05; **p<0.01; ***p<0.001.
Informed consent was granted by the parents, and the study was approved by the Hadassah ethical review committee. All heterologous expression experiments were approved by the Tel Aviv University Institutional Animal Care and Use Committee.
Results and discussion
The subjects were three sibs affected with a severe neurodevelopmental disorder of early infancy characterised by intractable epilepsy and lack of acquisition of any developmental milestones. Because of the parental consanguinity and the patients of both sexes, we assumed a founder mutation transmitted in an autosomal recessive manner. In order to identify the disease-causing mutation we searched for common homozygous regions in the patients’ DNA which were not shared by their unaffected sister. This analysis, performed with DNA SNP array, resulted in the identification of only two regions larger than 2 Mb, chr3:43071923-60527479 and chr7:149752241-154181631 (numbering according to assembly HG19). The regions, spanning a total of 21.89 Mb, encompassed 277 protein coding genes, consisting of 3172 exons. Because of the large number of exons, we opted for whole exome sequencing in the DNA sample of one of the patients. Filtering of the called variants retained 204 homozygous exonic variants which were not present in dbSNP130 and had a reading depth of at least ×7. Nonetheless, within the two homozygous regions there were only six non-synonymous variants and only two of those were predicted to be pathogenic by SIFT (Sorting intolerant From ntolerant) and Mutation Taster software. These changes were on chr3: 48682550C->T (rs149614835) causing Met2630Ile in the CELSR3 protein; and on chr3:50402595 A->G resulting in c.3119 A->G (p.Leu1040Pro, L1040P) at the CACNA2D2 gene. In mice, inactivation of CELSR3 was associated with severe malformations of the forebrain,8 whereas a large in-frame insertion in CACNA2D2 was associated with an EEG recording indicative of absence epilepsy.9 In view of the normal gross anatomy of the brain in our patients and the predominant seizure disorder, we focused on the CACNA2D2 mutation. The mutation segregated with the disease in the family (figure 1A–D) and affected a highly conserved codon—primates, rodents and fish all had Leu or Ile at this position, and the three other CACNA2D genes also had Leu or Ile at this position. The mutation was not present in 102 ethnically matched controls or in dbSNP135, nor was it found in 6503 healthy individuals whose Exome analysis results are available through the Exome Variant Server, NHLBI Exome Sequencing Project, Seattle, Washington, USA (http://evs.gs.washington.edu/EVS/) (July 2012).
The CACNA2D2 gene encodes the α2δ2 auxiliary subunit of high voltage gated calcium channels (hVGCC). hVGCC are heteromultimeric protein complexes composed of the main channel forming α1 subunit, which carries calcium influx across the plasma membrane, and the auxiliary subunits β, γ, and α2δ. The auxiliary subunits modulate calcium currents and channel activation and inactivation kinetics.10 ,11 These subunits are also involved in the proper assembly and membrane localisation of the calcium channel complexes.12 ,13 α2δ is a receptor for the antiepileptic drug gabapentin,14 ,15 and is associated with the enhancement of hVGCC expression and current amplitudes.10 ,16 ,17 The α2δ2 subunit is post-translationally cleaved into a long extracellular α2 protein and a shorter membrane anchored δ polypeptide;18 the α2 and δ proteins are linked by disulphide bonds.19
In order to study the consequences of the L1040P mutation in the CACNA2D2 gene on the properties of channels involved in neural signalling, we expressed the wild-type (WT) and the mutated CACNA2D2 gene in Xenopus oocytes and monitored their effects on N (CaV2.2) and L (CaV1.2) type calcium channels. When α2δ2-L1040P was expressed instead of α2δ2-WT, Ba2+ currents of CaV2.2 (α1B+β3+α2δ2) were reduced by >65% and were comparable to currents of α2δ2-less channels (figure 3A, B left). A decrease of 85% was also observed in the CaV1.2 current (figure 3B, right). CaV2.2 channels expressing α2δ2-L1040P also showed a slower time dependent current decay (figure 3A). This decay reflects the inactivation process, a negative feedback mechanism that prevents an excessive entry of Ca2+ to the cell. Inactivation in hVGCCs is divided into voltage dependent inactivation (VDI) observed with Ba2+ as the charge carrier, and Ca2+ dependent inactivation (CDI) which occurs, along with VDI, with Ca2+ as the permeable ion.11 The extent of inactivation was quantitated as the fraction of current remaining after 400 ms of depolarisation, r40020—the higher the r400, the weaker the inactivation. CaV2.2 channels expressing α2δ2-L1040P showed attenuated VDI (figure 3A, C) and CDI (figure 3D) compared to α2δ2-WT; the VDI became as slow as in channels expressed without α2δ subunit (figure 3C, bottom curve). We next expressed α2δ2-L1040P on top of α2δ2-WT, along with α1B and β3 subunits. VDI was strongly accelerated (low r400) by α2δ2-WT, and expression of even a fivefold excess of RNA of α2δ2-L1040P did not alter r400 significantly (figure 4), indicating that α2δ2-L1040P lacks a dominant negative effect, in correlation with the absence of epileptic symptoms in the healthy heterozygous family members.
To understand further the mechanism by which the mutation in α2δ2 changes current properties, we checked whether the mutated protein was properly synthesised in the cell. We detected the α2δ2 protein using a specific antibody, in cell lysates of oocytes expressing α1B+β3 without α2δ2 WT, with α2δ2-WT or with α2δ2-L1040P. In addition to an endogenous α2δ2 present in native oocytes,21 we detected a ∼170 kDa band, a size that corresponds to purified α2δ2 as it runs on SDS-PAGE, in the group expressing WT or mutated α2δ2 (figure 5A). We concluded that the α2δ2-L1040P protein is synthesised in the oocytes. To investigate whether α2δ2-L1040P alters the trafficking of α1B to the plasma membrane, we used a fluorescently (YFP) labelled α1B, and expressed it with β3 subunit without α2δ2 or with either WT or mutant α2δ2. We found that YFP-α1B fluorescence signal in plasma membrane decreased by 65% when expressing α2δ2-L1040P, compared to the WT α2δ2 (figure 5B). A similar decrease in fluorescence was measured when the channel was expressed without α2δ2. This result demonstrates that the inability of α2δ2-L1040P to increase the plasma membrane expression of CaV2.2, and/or to associate with α1, is the basis for its dysfunction.
Our results suggest that L1040P mutation in the α2δ2 protein is associated with dysfunction of α2δ2 rather than a decrease in its synthesis or a dominant negative effect of the synthesised protein. The potential involvement of the mutated α2δ2 protein in the epileptic phenotype may rely on retention of the α1 subunit in the endoplasmic reticulum and/or an improper assembly of the channel, generating a channel lacking the α2δ2 subunit in the neurone's plasma membrane. α2δ2-L1040P is unable to reproduce the most common effects of α2δ2-WT: an increase in whole cell amplitude and acceleration of inactivation.22 ,23 Since α2δ increases hVGCC currents both via effects on channel gating and surface expression,6 ,24 ,25 the α2δ2-L1040P mutant appears deficient in performing one or both of these essential functions. Prolonged channel opening during depolarisation, associated with slow inactivation, could partly underlie the epileptic phenotype; low current amplitude would add to the development of an aberrant calcium signalling pattern. Furthermore, lack or dysfunction of α2δ2 is expected to affect synaptogenesis and spatio-temporal calcium signalling,13 ,25 aggravating the pathology. Our clinical data and the results of the electrophysiological studies are in agreement with the spike wave seizures reported in mice carrying a spontaneous mutation in α2δ2 (ducky alleles) associated with an aberrant α2δ2 protein.26–29 We cannot exclude the possibility that our patients’ phenotype is aggravated by the concomitant occurrence of the homozygous Met2630Ile mutation in the CELSR3 protein; this rare mutation affects a conserved codon and the protein was shown to play a role in the connectivity of the cerebral cortex.30
Human voltage gated calcium channelopathies with primary central nervous system manifestations have hitherto been solely reported in association with α1A, the principal subunit of the voltage gated P/Q-type calcium channel encoded by CACNA1A. Autosomal dominant mutations in this gene were clinically manifested by episodic and progressive forms of cerebellar ataxia (EA2 and SCA6), familial hemiplegic migraine (FHM1), vertigo, and epilepsy.31 The pathophysiology of cerebellar atrophy in patients with mutations in CACNA1A or CACNA2D2 remains obscure, but may perhaps serve as a neuroimaging marker for this newly emerging group of voltage gated calcium channelopathies.
The excellent technical assistance of Rachel Dahan, Noa Cohen and Lital Sheva is acknowledged.
SE and SO contributed equally
Contributors SE, ND, OE conceived and designed the experiments; SO, AS, SZ performed the experiments; SE, SO, AS, ND, OE analysed the data; SE, SO, ND, OE wrote the paper; SE, FAA, FBT undertook patient management, collection of samples, and delineation of the phenotype.
Funding This work was supported in part by the Joint Research Fund of the Hebrew University and Hadassah Medical Organization to Simon Edvardson.
Competing interests None.
Ethics approval Hadassah ethical review committee and Tel Aviv University Institutional Animal Care and Use Committee.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement The DNA SNP chip data and the exome data are available upon request.
Informed Consent Obtained.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.