Episodic Ataxias: Faux or Real?

The term Episodic Ataxias (EA) was originally used for a few autosomal dominant diseases, characterized by attacks of cerebellar dysfunction of variable duration and frequency, often accompanied by other ictal and interictal signs. The original group subsequently grew to include other very rare EAs, frequently reported in single families, for some of which no responsible gene was found. The clinical spectrum of these diseases has been enormously amplified over time. In addition, episodes of ataxia have been described as phenotypic variants in the context of several different disorders. The whole group is somewhat confused, since a strong evidence linking the mutation to a given phenotype has not always been established. In this review we will collect and examine all instances of ataxia episodes reported so far, emphasizing those for which the pathophysiology and the clinical spectrum is best defined.


Introduction
Episodic Ataxias (EA) are a genetically heterogeneous group of autosomal dominant disorders characterized by attacks of movement incoordination (cerebellar ataxia) of variable duration and frequency, often accompanied by additional ictal and interictal symptoms. The two most frequent types are EA1 and 2, respectively produced by mutations of genes altering the function of K v 1.1 and Ca v 2.1 ion channel. Both diseases have a wide spectrum of clinical manifestations. Mutations of the CACNA1A gene, coding for the alpha subunit of Ca v 2.1 channels, are the cause of two other allelic disorders, Spinocerebellar ataxia 6 (SCA6) and Familial Hemiplegic Migraine 1 (FHM1). These have features that can overlap with those of EA2. EA3 to 8 are either very rare or reported in single families. For only two of them, EA5 and EA6, the mutated gene is known, while for the others no responsible gene has been identified, except for EA8 possibly associated with two candidate genes (for references see below). In all of them, episodes of ataxia of variable frequency and duration are the most prominent and constant sign, even if sometime associated with other abnormalities (see below). EAs, however, should be distinguished from other genetic disorders in which episodes of ataxia are not the main clinical feature, but just a possible phenotypic variant.
There is not always strong evidence for the progressive increase in the number of EAs and for the expansion of their clinical spectrum (including both the group of EA1-8 and the disorders with a variant EA phenotype). Sometimes, the association between the gene mutation and the clinical picture is not confirmed by functional analyses, or other hypotheses are possible. In this review, we will describe the whole panorama of EAs, emphasizing those for which the pathophysiology and the clinical spectrum is best defined. phenotypes (EA2 and EIEE) within these families confirms that EIEE is part of the EA2 spectrum. De novo missense mutations of CACNA1A underlying EEIE cases have also been reported [42][43][44], mostly without a functional mutation analysis. Jiang et al. [38] performed a functional analysis of some de novo missense mutations causing EIEE which surprisingly revealed that some of them are loss-and other gain-of-function mutations. A severe epileptic phenotype was also present in two siblings with a heterozygous compound CACNA1A mutation (a missense and a 7 amino acid frameshift deletion), whose heterozygous relatives had only mild intellectual deficit, but no episodes of ataxia [45]. The phenotype of the two sibs also included features never previously reported in EIEE due to CACNA1A mutations, such as dysmorphisms and optic atrophy, leading to complete blindness. The unusual phenotype could be due to the combined mutations, but the presence of a chromosomal rearrangement in one of the parents, leading to a microdeletion in offspring, cannot be ruled out, as it would not be revealed by Whole Exome Sequencing.
Atypical phenotypes have also been reported in chromosomal microdeletion carriers, including part of (or the whole) CACNA1A sequence as well as other genes, but no assessment was made about the possible influence on phenotype of chromosomal imbalance, or of other gene deletions [39,46].
The spectrum of EA1 and EA2 phenotype has been progressively broadened, but the underlying mutation has not always been investigated through functional analyses, leaving some uncertainties about their effects, particularly when a novel/de novo mutation was found with a pathogenicity assessed only in silico. There is no doubt, however, that both EA1 and EA2 have an extremely variable clinical phenotype. There can be various reasons for this variability. It has been hypothesized that different mutations exert different loss-of-function effects or modify the function of the proteins that interact with the two alpha subunits [1,3,5,51]. Furthermore, the widely different clinical features in different members of the same family, sharing an identical genotype, [40,52,53] suggests the influence of modifiers in the genetic background. Moreover, clinical differences in monozygotic twins [54,55] suggest the influence of environmental factors. In any case, such a wide interfamilial and intrafamilial phenotypic variability makes it extremely difficult to define genotype-phenotype correlations [56].
The preventive management of both disorders mostly involves avoiding triggering stimuli that patients usually know quite well. Pharmacologically, little can be done to restore the channel loss-of-function. In EA2, however, acetazolamide (AAA)-a carbonic anhydrase inhibitor-can reduce the frequency of attacks, or completely abolish them [57]. So far, no placebo-controlled trials have been performed to assess the efficacy of AAA [58], and its therapeutic mechanism of action is not well understood. Magnetic resonance spectroscopy in EA2 patients showed increased pH in the cerebellum and cerebrum, which was corrected upon AAA administration [59]. Because of pH adjustments, channels and ionic conductance across neuronal membranes may be modulated, causing membrane hyperpolarization and a decrease in excitability, which can result in the observed reduction of attacks [1]. AAA, however, can have unwanted side effects such as kidney stones, hyperhidrosis, paresthesia, muscle stiffening with easy fatigability, and gastrointestinal disturbances [60], and it cannot always be prescribed.
In a pilot study and in an experimental trial, 4-Aminopyridine (4-AP)-a K + channel blocker-was effective in reducing the frequency of EA2 episodes and improving patient wellbeing [61,62]. 4-AP acts on Purkinje cells, by ameliorating their pace-making activities [63,64].
In EA1, a number of drugs have been used with variable therapeutic effects [1,65]. AAA has been shown to be only occasionally effective in EA1 [66,67]. Considering the increased relative risk of seizures in these patients, various antiepileptic medications have been used with variable results, such as phenytoin [68][69][70], carbamazepine [8,32], and lamotrigine [16]. It should be noted, however, that over 50% of patients have never tried any preventive medication [16]. Very recently, small molecules selectively opening Kv1.1 channels have been investigated, thus offering a possible therapeutic drug for EA1 [71]. Moreover, in a recent preclinical study using a rat model of focal neocortical epilepsy gene, the overexpression of Kv1.1 was effective in controlling seizures [72], thus potentially offering the prospect for a targeted correction of the genetic defect in very severe EA1 cases. (Table 1) EA5 was reported as affecting one in 71 analyzed EA families. The family carried a p.C104F missense mutation in CACNB4 gene coding for the beta 4 auxiliary subunit of Ca v 2.1 channels [73]. The same mutation was also present in an unrelated family with a generalized epilepsy and no episodic ataxia [72]. Functional studies in vitro showed no alteration of channel kinetics, but the authors hypothesized other pathogenic effects not revealed by the analysis on channel gating [73]. This variant, although very rare, is present in the general population (Genome Aggregation Database and Ensembl). It is well known that CACNB4 mutations are causing different types of epilepsy [65]. No other cases presenting with episodes of ataxia, however, have been so far reported, despite the inclusion of CACNB4 gene among those screened for mutations in EA patients [74,75]. Jen and Wan [13] have called into question whether there are sufficient data to support the designation of EA5. ?? indicate uncertainties described in the text.

EA5 and EA6
EA6 patients carry mutations in the excitatory amino acid transporter 1 (EAAT1), a Na + -dependent glutamate transporter coded by gene SLC1A3 and expressed in cerebellar astrocytes known as Bergmann glia [75][76][77][78]. The patients present with EA as well as progressive ataxia, seizures, and migraine headaches with [76] or without [77] prolonged alternating hemiplegia, triggered by head trauma or fever. The functional study of the mutations showed a loss of transporter function with impaired glutamate uptake. EAAT1, however, also has a function as ion channel; the same mutation p.P290R reported by Jen et al. [76] was found to induce a gain of this second type of function. It has been proposed that a similar mechanism is the primary cause of EA6, rather than the reduction of glutamate uptake [79,80]. A recent variant of EAAT1 carried by a patient presenting with a severe migraneous headache was shown to impair K+ binding to the mutant protein [81], thus introducing a novel mechanism causing glutamate transport dysfunction.  (Table 1) Each of these EAs, all AD, has been reported in one or two large families presenting with a phenotype largely overlapping with that of EA1 or EA2, except for a few clinical signs. EA3 had short attacks with ictal and interictal signs similar to EA1 except for vertigo and tinnitus [82]. EA4 was similar to EA2 except for a late onset [83]. EA7 was similar to EA2 except for absence of interictal nystagmus [85]. Attacks in EA8 patients were similar to EA2, but were not sensitive to AAA, while responding to clonazepam [86].
A genome-wide linkage analysis in the EA3 family reported by Steckley et al. [82] mapped the gene in a 4-cM region on 1q42 with a relatively high lod score. However, this result was obtained only after adapting the linkage parameters to the family. This was done by considering three affected patients that did not carry the putative haplotype as phenocopies, and by considering five healthy family members with the disease haplotype as instances of incomplete penetrance. Otherwise, the lod score would have been below the significance cutoff [87].
In a 91-year-old patient of the EA4 family [83], a neuropathological brain examination showed, among other abnormalities, the presence of poliglutamine repeats in Purkinje and granule cells, without intranuclear inclusions, similar to those of SCA6 brains [88].
EA7 was mapped in a 10-centimorgan on chromosome19q13 with a lod score slightly above the significance cutoff [85]. Sequencing of two candidate genes KCNC3 and SLC17A7, lying in this region, did not reveal any mutation. So far, no other EA3, EA4, or EA7 families have been reported, making it possible that gene mapping results are not sufficiently reliable, or that the disorder was, in fact, one of the known types of fluctuating ataxia, such as SCA6.
EA8 was mapped in a relatively large region of 18.5 Mb on chr. 1p36.13-p34.3 with a lod score very near to cutoff [86]. Exome sequencing in the latter region revealed a variant in two genes SPG2 and UBR4, the pathogenicity of which was predicted in silico to be stronger in the second gene than in the first. UBR4 is an ubiquitin ligase protein that is interacting with calmodulin and co-localizes with ITPR1 with a potential for disrupted Ca ++ control within neuronal cells [86]. Two additional EA cases with a UBR4 mutation have been reported [75], but, as in the previous case, no functional analysis was performed.

Episodiac Ataxias in the Context of Other Disorders
Some of EA-like phenotypes have been reported in the context of other genetic disorders, either as the most prominent feature or in association with other abnormalities. This could explain, at least in part, the presence of patients with typical EA1 or EA2 clinical picture who do not carry mutations in the respective genes [16,17,74,89]. Table 2 shows an updated list of these disorders. Some of them are related to mutations in genes coding for ion channels (SCNA2. KCNA2, KCND3, NALCN) or ion pumps (ATP1A2 and ATP1A3), or proteins interacting with ion channels (PRRT2 and FGF14) or Na + -dependent transporters (SLC2A1). For most disorders in this group, the EA variant phenotype is reported in a single case, and often associated with other abnormalities (see references in Table 2). For PRRT2 and FGF14, and SCNA2, instead, the inclusion of EA in the disease spectrum was confirmed in several patients. PRRT2 (Proline-rich Transmembrane protein 2) mutations underlie three major phenotypes: Benign Familial Infantile Epilepsy, Paroxysmal Kinesigenic Dyskinesia with or without infantile convulsions, and Infantile Convulsions with Choreoathetosis. About 5% of patients bearing PRRT2 mutations display other disorders such as episodic ataxia, hemiplegic migraine, developmental delay, and intellectual disability [99][100][101][102]. The pleiotropy associated with PRRT2 mutations is not related to any specific genotype-phenotype correlation, as most mutations are loss-of-function [103]. PRRT2 is a neuron synaptic protein expressed at the highest levels in cerebellum, basal ganglia, and neocortex. At the nerve terminal, PRRT2 endows synaptic vesicle exocytosis with Ca ++ sensitivity by interacting with Ca ++ -sensing machinery, thus playing an important role in calcium triggered exocytosis [103][104][105]. PRRT2 knock-out mice show heightened spontaneous and evoked activity at the network level associated with increased excitability of excitatory neurons [105]. PRRT2 was also found to interact with Na + channels and their direct consequences on neuronal excitability [106]. Carbamazepine is reported to be an effective drug for PRRT2 mutations [99,102].

FGF14
Mutations in FGF14 gene, coding for Fibroblast Growth Factor 14, underlie SCA27, a late onset, slowly progressing cerebellar ataxia with extrapyramidal features such as postural tremor, head titubation, and parkinsonism [107,108]. This gene, highly expressed in the brain and especially in granule and Purkinje cells, controls channel gating and axonal targeting of Na v 1.1, 1.2 and 1.6 channels [109,110]. It also regulates Ca v 2.1 channels [111] and is required for Purkinje cell spontaneous firing [112]. Several patients have been recently reported with an autosomal dominant episodic ataxia harboring FGF14 mutation [74,[113][114][115][116][117]). Episodes have variable onset, frequency, and duration and are characterized by vertigo, dizziness, and unsteadiness often triggered by fever. Tremor and nystagmus are usually interictal signs. Responsiveness to AAA has been reported [116,117]). Piarroux et al. [117] suggest to consider FGF14-related episodic ataxia as Episodic Ataxia type 9.

SCNA2
SCN2A encodes the alpha subunit of the voltage gated neuronal Na V 1.2 channel. Loss-of-function mutations are associated with a severe form of epilepsy, and/or intellectual disability, and autistic traits. Gain-of-function pathogenic variants of SCN2A underlie the Benign Familial Infantile-Neonatal Seizure (BFNIS) affecting children before three months of age and disappearing with age. Seizures in BFNIS can be controlled with Na + -channel blockers such as phenytoin and carbamazepine, while they are ineffective for the severe form [118,119]. Some BFNIS patients present with EA either co-occurring with epileptic spells [120][121][122][123][124][125] or as the only clinical feature [124][125][126]. EA does not respond to Na + channel blockers, suggesting a distinct pathophysiological mechanism, which, however, has not been identified so far. Cases of EA responding to AAA has been reported [123,126].

Mitochondrial Disorders
Another group of diseases, for which single EA cases have been reported, is related to mitochondrial metabolism (PDHA1, TPK1, and DARS2 gene). A common trait of these diseases is the presence of CSF and plasma lactic acidosis, never found in EA1-8. In these cases, EA does not present as the prominent symptom, but is associated with other abnormalities.

Other Disorders
In a final group of highly heterogeneous disorders, the only disease with a prominent EA phenotype is GLUT1 deficiency syndrome, due to mutations in SLC2A1 gene on chromosome 1p34. The gene codes for the membrane protein responsible for glucose transport across the blood-brain barrier, maintaining the continuous high glucose and energy demands of the brain [127]. Mutations in this gene cause the Glucose Transporter 1 deficiency syndrome (GLUT1 DS) and reduce the maximum velocity of glucose uptake [128,129]. The phenotype is characterized by early onset epilepsy, mild to severe developmental delay, and acquired microcephaly. Hypoglycorrhachia and low CSF lactate are diagnostic symptoms. Epileptic attacks and other manifestations, such as developmental delay and intellectual disability, are often controlled by a ketogenic diet. Ketone bodies diffusing across the blood-brain barrier, facilitated by a monocarboxylic acid transporter, serve as an alternative energy source for brain metabolism [128]. About 10% of SLC2A1 mutation carriers do not have the classic epileptic phenotype and present with paroxysmal exertion-induced dyskinesia which, in some cases, is reported as EA [129]. Often patients presenting with EA also have other neurological abnormalities [130][131][132][133][134], but cases with an almost pure EA have been reported [129,135]. AAA in these cases controls ataxic spells [129,132,135].

Conclusions
EAs are a group of disorders with many complex aspects. A first complicating element is the high variability of phenotypes associated with different mutations in the same gene. Some of these could be related to the mutation functional effect (gain-or loss-of-function), as e.g., for CACNA1A or SCNA2 gene, but even within the same functional effect the phenotypes are highly variable. The site of the protein mutation can rarely predict the phenotype, as in EA1 patients carrying a p.T226R substitution who have quite different clinical features from those carrying p.T226A or p.Thr226M mutations [1]. Moreover, phenotypic variability was also reported among family members sharing an identical mutation [52][53][54] or in monozygotic twins [54]. The underlying genetic and pathophysiological basis of the extreme variability within the same disease has not been clarified, thus preventing any genotype/phenotype correlation. Genetic modifiers as well as environmental factors are likely to influence the clinical picture, but both have been poorly investigated so far. The recent possibility of analyzing channel function in iPSC derived neurons from patient fibroblast [136,137]-hence sharing the patient genetic background-could perhaps throw some light on this problem.
In some disorders, the broad extension of the clinical spectrum is documented in reports of single cases with an atypical phenotype. This leaves room for doubts about their real association with the reported mutation particularly in de novo cases-for which genotype/phenotype co-segregation cannot be assessed-and/or when gene variants are the result of NGS methods. The limits of these methods may hinder other mutations in sites not sufficiently covered or undetectable (e.g., repeat expansion, microdeletion). Moreover, more uncertainties are caused by the fact that in these cases no functional study of the mutated protein is usually performed.
In conclusion, it would be helpful to reconsider the whole nosography of EAs in the light of the present and future genetic and pathophysiological evidence.