De novo variants in CACNA1E found in patients with intellectual disability, developmental regression and social cognition deficit but no seizures

Voltage-gated calcium channels (VGCCs) are proteins that regulate the entry of calcium ions (Ca²⁺) in response to depolarization of the cellular membrane of excitable cells [1] and represent principal gateways for Ca²⁺ in nerve and both skeletal and cardiac muscle cells [2]. VGCCs are constituted by 5 different subunits, α1, α2, β, δ, γ, which together form the high-voltage-activated Ca²⁺ channels [3]. The α1 subunit is the main subunit indispensable for the functioning of the channel. It is composed of four homologous domains (I–IV), each containing 6 transmembrane segments, including a voltage sensor segment (S4) and 2 segments forming a pore (S5–S6) [1]. In mammals, ten α1 subunits have been described, encoded by ten different genes. Each of these subunits has specific molecular, genetic, physiologic and pharmacological properties [1, 4], and pathogenic variants in these genes have been associated with various human diseases and represent putative targets for treatment [5].

The α1E gene (CACNA1E) encodes the high-voltage-activated Cav2.3 type R calcium channel, which is expressed in various areas of the central nervous system, including the cerebellum [1]. More specifically, it is involved in the initiation of the presynaptic calcium entry and post-synaptic transmitter release [6].

CACNA1E variants were initially associated with seizures (Baker JJ 2016: http://​epostersonline.​com/​acmg2016/​node/​2302) and with autism spectrum disorder (ASD) [7]. Other studies found a possible impact of CACNA1E single nucleotide polymorphisms in different human diseases and disorders including diabetes [8], migraine [9] and pain regulation [10]. More recently, the pathogenic role of CACNA1E variants in human epilepsy and neurodevelopmental disorders (NDD) has been described. One large genomic study of trios presenting with NDD identified CACNA1E as a candidate gene for NDD with or without epilepsy [11], and subsequently, variants in CACNA1E have been found in patients with developmental and epileptic encephalopathy (DEE) [12]. A large multiplex gene network analysis identified CACNA1E as candidate gene for epilepsy and autism [13]. Subsequently, a genetic association between calcium channels in general and ASD was further suggested by Liao and Li [14] and Graziano et al. [15].

The largest cohort of CACNA1E patients, described by Helbig et al. [12], reports 30 patients, aged between 1 and 16 years, who presented with severe to profound developmental delay and a combination of epilepsy, tone disorders and/or abnormal movements. Among these individuals, 88% were described as non-verbal and non-ambulatory. Of the 30 patients, 29 patients presented with hypotonia, six presented with spastic quadriplegia, 13 with joint contractures, 14 with extrapyramidal signs, and 26 with epilepsy (among these, 14 had epileptic spasms). Nine patients presented with developmental regression associated with the onset of epilepsy [12]. Interestingly, topiramate was the main anti-seizure drug reported to show a reduction in seizure presentation. Functional analysis of the Cav2.3 channel in 14 patients revealed a gain-of-function effect of the CACNA1E pathogenic variants with either a facilitated activation or an increased Ca²⁺ current. As topiramate blocks R-type calcium channels, among its different actions, the R-type calcium current might represent a common neuropathogenic mechanism involved in epilepsy and developmental delay [16].

We identified a de novo missense variant in CACNA1E in a patient with global developmental delay (GDD), ASD and successive developmental regressions unrelated to epileptic seizures. We used GeneMatcher [17] to seek other similar developmental patterns associated with CACNA1E variants, with the aim to identify possible neurodevelopmental presentations without epilepsy and to extend the phenotypic description of CACNA1E-related disorders. Using large datasets of control populations, we tried to compute the mutational constraint within CACNA1E domains and segments to identify putative functionally relevant segments that are more likely to harbor pathogenic variants.

The seven individuals (six children and one young adult) with heterozygous de novo pathogenic variants in CACNA1E described here showed a broad phenotypic spectrum including intellectual disability, GDD, abnormal behavioral phenotypes (in three cases associated with clinically confirmed ASD diagnosis), hypotonia, absence of language or speech delay and developmental regressions in infancy, but no severe neuromotor deficit or epileptic seizures (except for epileptic spasms in Patient 4). This cohort has enabled us to broaden the genotypic and the phenotypic spectrum of CACNA1E-related neurodevelopmental disorder.

The phenotype in our patients was milder than the one reported previously [12], which might be linked to the absence of refractory epilepsy. Indeed, the 30 patients previously reported [12] presented with a more severe neurodevelopmental phenotype, as the expression of a more severely impacted brain developmental trajectory. Among them, four had no epilepsy and one had no information available for the epilepsy phenotype. The non-epileptic patients older than 2 years (2/4, 50%) showed a milder phenotype than the others in the cohort, having single-word capacity and being ambulant. Cognitive and behavioral profiles reported for those patients were very similar to the ones seen in our cohort. However, social deficits were not reported in their study.

Pathogenic CACNA1E variants reported previously [12] and in this paper cluster mostly inside or at the end of segment 6 of Domains I, II and III (Fig. 1A). Based on the analysis of genetic intolerance, those regions are the most intolerant to missense variation (Fig. 2). Thus, the natural variant epidemiology supports the concept that these regions are most crucial for the normal function of CACNA1E [16, 24].

Based on the clinical description of the patients harboring the different CACNA1E variants, there seems to be no clear link between the localization of the variant on the protein and the clinical outcome, especially of refractory epilepsy or spastic quadriplegia (Fig. 1B). Indeed, four of the patients presented here carry missense variants impacting either the same amino acid (p.Ala702) or in the same clustered region of the segment 6 in Domain II as some patients reported with severe epilepsy [12]. Interestingly, among the patients without epilepsy reported previously, two harbor missense variants also seen in patients with severe epilepsy [12]. The first patient is a 1-year-old girl with a p.Gly352Arg missense variant, a recurrent mutation seen in 8 other patients with a strong epilepsy phenotype, which started after age 1 in 3/8 cases. The second non-epileptic patient is a 6-year-old boy carrying a p.Thr1425Asn variant, also seen in a second patient who had epileptic spasms at 5 months of age, but none since then.

Based on the existing reports as well as the case series reported here, pathogenic variants in CACNA1E appear to result in a disease with a spectrum of clinical manifestations ranging from severe refractory epilepsy with globally, severely disturbed development (DEE) to GDD with intellectual disability, developmental regression and ASD. This high phenotypic heterogeneity is not uncommon in both adult and developmental channelopathies [25], and it is made even more evident by the observation of different phenotypes resulting from identical mutations. Indeed, it is likely that other mechanisms including modifier genes, environmental factors or variations in gene expressivity contribute to the variability of the phenotypes. Considering the variability in the age of epileptic onset in the cohort of Helbig et al. [12], ranging from 1 day to 3 years, we cannot exclude that the younger patients in our cohort may develop seizures later in life.

The identification of a splicing variant adds further complexity to the pathogenic mechanisms. Our patient 7 has a variant predicted to disrupt exon 22 consensus splicing donor site (c.3422+1G>A). Helbig et al. have tested electrophysiologically the functional effects of the three CACNA1E missense variants in the segment 6 of Domain II and one variant located in the S4-S5 intersegment of Domain II. These 4 missense variants all led to an increase in current density, facilitated channel activation and slower deactivation, and were thus consistent with gain-of-function effect of the variants. Additionally, the study also described three loss-of-function variants impacting CACNA1E and suggested that haploinsufficiency might be another pathogenic mechanism. The phenotypes of those three patients were milder than the others in their study and resemble that of the patients in our series. The presence of both haploinsufficiency variants as well as missense variants that may act through dominant negative or perhaps even gain-of-function mechanism is intriguing, but not novel. The existence of both loss-of-function and gain-of-function variants in calcium channel genes has been reported in CACNA1A for variants leading to variable severity of DEE [26], as well as in other neuronal channels (SCN8A [27], KCNQ2 [28], KCNQ5 [29], KCNA2 [30], KCNB1 [31]). Interestingly, gain-of-function changes in the electrophysiology of the cell can also result from loss-of-function effects of variants at the molecular level, and therefore, the gain-of-function effect of truncating and/or splicing variants in CACNA1E should not be excluded without further functional studies of these variants. Based on the genetic epidemiology of variants in gnomAD and our computations, the CACNA1E gene seems to be constrained both against missense as well as against loss-of-function variants. Thus, the pathogenic mechanism remains unclear and this complicates genotype–phenotype correlation analysis. Such molecular understanding would be therefore needed in the future in order to design therapeutic approaches with drugs such as topiramate.

Furthermore, the observations in this small cohort challenge our current understanding of the interplay between epilepsy and its impact versus its co-occurrence with developmental delay. The term DEE was coined by the most recent classification from the International League Against Epilepsy (ILAE), defining relationship between developmental delay with or without early-life pharmacoresistant epilepsy [32]. However, recent publications of subjects harboring pathogenic variants in so-called epilepsy genes who did not have seizures suggest that the concept of variants specific to epilepsy in the context of syndromic presentation may be mistaken. In that aspect, epilepsy, or ASD, might be considered as one of the phenotypic expressions of a monogenic syndrome [33]. The present work adds CACNA1E to the expanding list of genes implicated in an autosomal dominant neurodevelopmental disorder, including ASD, with or without epilepsy. Growing evidence suggests that diagnostic gene panels for epilepsy should be merged with larger brain development gene panels.

In Patient 1, treatment of topiramate [12] was introduced with the hypothesis that modulation of the dysfunctional α1 subunit might improve the function of the neuronal connectivity underlying behavioral disorders, independently of the electroencephalographic activity as the patient never presented with any seizure. Parents reported a qualitative improvement concerning sleep quality and sphincter control; however, it might had happened by chance or as a natural evolution of the disorder. No additional effect has been observed on his behavioral disorder and development.

This study provides a further example of how matchmaking tools like GeneMatcher can be used to define expanded phenotypes associated with described disease genes, like it has been previously described for LMNB1 [34] and GATAD2B [35]. As NGS analysis is becoming routine worldwide and many laboratories are faced with variants that may not correspond to “typical” phenotypes, these sharing and matchmaking tools will become even more important to facilitate the recognition of novel phenotypic associations and variant interpretation.

The main limitation of our study is the absence of functional studies of the effects and impacts of the identified variants on Cav2.3 channel function. Due to this lack of experimental data, we cannot assess whether the variants reported have a gain-of-function effect, as described in Helbig et al. [12]. Considering the similar types of variants (missense) and the consistent NDD phenotype, the de novo origin of these variants, their absence in control population, their impact on conserved amino acid and deleterious predictions and the presence of nearby pathogenic variants, our reported variants are classified as likely pathogenic/pathogenic according to ACMG criteria [18]. Functional experiments are nevertheless essential to understand the pathomechanism behind the reported variants.

The second limitation of our study is the small cohort size. Gathering of patients through matchmaking platforms comes with clear benefits but at the same time is limited by its multi-centric nature, which leads to involvement of variable specialists and assessment tools for each patient. Similarly, the individual submitting the information on the matchmaking platforms might not always be in direct contact with the patients and their referring clinicians, explaining the variability in clinical assessments and detailed history of the patients, for example here the presence/absence of early regression or diagnostic criteria and full assessment for ASD. It thus shows how crucial the collaboration between clinicians and molecular specialists might be in the future.

Topiramate was tested in only one patient of the cohort, and no specific recommendations can be made, as the indication for its use was speculative. However, this information might support the use of this anti-seizure medication drug in unusual indications.

In addition, our study highlights the role of both splicing loss-of-function and missense variants but does not elucidate the underlying molecular mechanisms.

In this study, we identified heterozygous pathogenic or likely pathogenic de novo variants impacting CACNA1E in seven patients presented with a phenotype that was less severe than that previously reported, notably with the absence of severe epilepsy, of macrocephaly and of spastic quadriplegia. The overlapping phenotypes in our patients include GDD, abnormal social behavior including ASD, hypotonia and developmental regression. Furthermore, we analyzed the variant landscape of CACNA1E and showed strong constraint of missense variants in specific segments of CACNA1E. We conclude that CACNA1E variants have a broader phenotypic spectrum than previously reported, including non-epileptic ASD and GDD, and this may inform the composition of diagnostic gene panels and the interpretation of variants observed in NGS studies.