Role of PRRT2 in common paroxysmal neurological disorders: a gene with remarkable pleiotropy
- School of Pharmacy and Medical Sciences and Sansom Institute for Health Research, University of South Australia, Adelaide, Australia
- Correspondence to Dr Sarah E Heron, School of Pharmacy and Medical Sciences, University of South Australia, P4-47, City East Campus, GPO Box 2471, Adelaide, SA 5001, Australia;
- Received 7 November 2012
- Revised 18 December 2012
- Accepted 20 December 2012
- Published Online First 23 January 2013
Mutations in the gene PRRT2 encoding proline-rich transmembrane protein 2 have recently been identified as the cause of three clinical entities: benign familial infantile epilepsy (BFIE), infantile convulsions with choreoathetosis (ICCA) syndrome, and paroxysmal kinesigenic dyskinesia (PKD). Patients with ICCA have both BFIE and PKD and families with ICCA may contain individuals who exhibit all three phenotypes. These three phenotypes were all mapped by linkage analyses to the pericentromeric region of chromosome 16, and were hypothesised to have the same genetic basis due to the co-occurrence of the disorders in some families. Despite considerable effort, the gene or genes for BFIE, ICCA, and PKD were not identified for many years after the linkage region was identified. Mutations in the gene PRRT2 were identified in several Chinese families with PKD, suggesting that the gene may also be responsible for ICCA and BFIE in families linked to the chromosome 16 locus. This was demonstrated to be the case, with the majority of families with ICCA and BFIE found to have PRRT2 mutations. The vast majority of these mutations are truncating and are predicted to lead to haploinsufficiency. PRRT2 is a largely uncharacterised protein. It is expressed in the brain and has been demonstrated to interact with SNAP-25, a component of the molecular machinery involved in the release of neurotransmitters at the presynaptic membrane. Therefore, the PRRT2 protein may play a role in this process. However, the molecular mechanisms underlying the remarkable pleiotropy associated with PRRT2 mutations have still to be determined.
Benign familial infantile epilepsy (BFIE), previously termed benign familial infantile seizures or benign familial infantile convulsions, is a self-limiting seizure disorder of infancy. Patients have non-febrile seizures with onset at between four and 12 months of age and offset by 2 years of age. Subsequent neurological development is usually normal. BFIE was originally described by Watanabe and colleagues1 in 1987 and the phenotype was further described and named ‘benign familial infantile convulsions’ by Vigevano and colleagues2 in 1992.
Watanabe and colleagues1 described nine infants with benign complex partial epilepsies characterised by the presence of clusters of seizures with motor arrest, decreased responsiveness, and automatisms. They observed that these infants had an apparently normal developmental outcome and that their seizures were easily controlled with antiepileptic drugs. This was in contrast to many cases of seizures in infancy, which were associated with an unfavourable outcome and developmental delay. Watanabe and colleagues1 also observed that four of the nine patients had a family history of benign infantile convulsions.
Vigevano and colleagues2 described a further five infants with clusters of partial seizures with secondary generalisation occurring between the ages of 4 and 7 months. The epilepsy was familial in all cases and had a favourable outcome. It was observed that the clinical features in these patients were similar to those seen in benign familial neonatal convulsions, apart from the age of onset. It was also observed that the features in these patients overlapped those described by Watanabe and colleagues.1 Vigevano and colleagues2 proposed the name ‘benign familial infantile convulsions’ for the disorder. This nomenclature has since been updated and the disorder is termed BFIE in the most recent classifications of the International League Against Epilepsy.3
Paroxysmal kinesigenic dyskinesia (PKD), also called paroxysmal kinesigenic choreoathetosis, is a movement disorder characterised by sudden attacks of involuntary movement that are induced by a sudden movement from rest, such as rising from a chair or starting to walk, or by exercise. The attacks in PKD consist of dystonic posturing, chorea or athetosis.4 Diagnostic criteria for PKD were proposed by Bruno and colleagues5 and include attacks with an identified kinesigenic trigger, short duration (<1 min), no associated loss of consciousness or pain, normal neurologic examination, exclusion of other causes, and onset at between 1–20 years of age or a family history of PKD. The disorder was delineated in detail in 1967 by Kertesz,6 who described the phenotypes of 10 patients from six families and reviewed similar reports from the literature. It was observed in this study that the disorder was often familial and therefore a genetic cause was proposed. PKD usually has onset in late childhood or adolescence and may remit in adulthood. The disorder responds well to treatment with antiepileptic drugs, particularly carbamazepine or phenytoin, and patients are otherwise normal.5–7 PKD shows autosomal dominant inheritance in families and sporadic cases are also observed.5 ,6
Infantile convulsions and choreoathetosis (ICCA) is a syndrome in which BFIE and PKD co-occur in the same patient or family. The syndrome was first described as a distinct clinical entity in 1997 by Szepetowski and colleagues,8 who identified four French families with autosomal dominant inheritance of BFIE and PKD. Genome-wide linkage analysis was performed for these families and all were found to be linked to a 10 cM interval in the pericentromeric region of chromosome 16. Following this report, additional families with ICCA and linkage to the same region were described.9 ,10 Families were also described with BFIE or PKD alone that showed linkage to the same region.11–19 ICCA, BFIE, and PKD were therefore hypothesised to be allelic.11 The minimal critical region (MCR) for the majority of these families corresponded to a 21.69 Mb (6 cM) region between D16S690 and D16S517 on chromosome 16 and was similar for ICCA, BFIE, and PKD (figure 1). A single BFIE family with a recombination event at D16S685,13 potentially reducing the MCR to a 2.7 Mb region between D16S690 and D16S685, and a second PKD locus on the q-arm of chromosome 1620 were also described (figure 1).
Identification of mutations in PRRT2
Despite the identification of the chromosome 16 BFIE/ICCA/PKD locus in 1997 and the subsequent extensive sequencing of candidate genes within the region,19 ,21 the causative gene was not identified for many years. In 2011, Chen and colleagues4 successfully employed a strategy combining linkage analysis and whole exome sequencing to identify mutations in an uncharacterised gene, PRRT2, in eight Chinese families with PKD. This finding was rapidly followed by many similar reports of mutations in the same gene in families from different ethnic backgrounds with PKD, ICCA, and BFIE.22–54 To date, over 330 families and individuals with mutations in PRRT2 have been described. De novo mutations are observed in sporadic cases of PKD, ICCA, and BFIE. The vast majority of these families (over 80%) have the same recurrent frameshift mutation: PRRT2 c.649-650insC; p.Arg217Profs*7. The remaining mutations in PRRT2 are spread throughout the gene (table 1, figure 2). Most of them are nonsense, frameshift and splice site mutations predicted to lead to protein truncation. Several of these mutations have been demonstrated to cause altered cellular localisation of the PRRT2 protein or loss of detectable protein expression in vitro.4 ,28 Fifteen different missense mutations have been reported.22–25 ,29 ,34 ,35 ,37 ,42 ,45 ,46 ,50 ,54 These alter amino acid residues clustered in and around two putative transmembrane domains located near the C-terminus of the protein (table 1, figure 2A). In addition to these substitution and small insertion or deletion mutations, three PKD or ICCA patients have been described with large sub-microscopic deletions encompassing a number of genes including PRRT2 (figure 2B).55–57 These findings indicate the need for copy number variant analysis as well as sequencing of PRRT2 to be included in a full diagnostic analysis of the gene.
Mutations in PRRT2 have also been identified in families with hemiplegic migraine (HM) and other forms of migraine. This association was initially described in a family with heterogeneous paroxysmal phenotypes including infantile seizures, PKD, HM, and paroxysmal torticollis.33 Migraine was also noted as a feature in some families where the primary phenotype was PKD, ICCA or BFIE.24 ,31 ,44 ,46 ,52 ,53 A small number of families with PRRT2 mutations have been described in which migraine, most commonly HM, is the only phenotype observed.44 ,46 In contrast to the high mutation rate observed in BFIE, ICCA, and PKD patients, PRRT2 mutations account for only a small proportion (0.7–3%) of cases of HM occurring without other paroxysmal disorders.44 ,45 ,47 The association of both migraine and seizure disorders with the same gene has been previously observed. A family with a mutation in ATP1A2 causing both migraine and infantile seizures has been described58 and mutations in SCN1A, which usually cause epilepsy, have been described in patients with HM.59 ,60
Occasional families with PRRT2 mutations have been described in which epilepsy phenotypes other than BFIE are observed.48 ,54 These phenotypes include febrile seizures, febrile seizures plus, and nocturnal convulsions. PRRT2 mutations have not been associated with other phenotypes that include infantile seizures. In particular, no mutations have been identified in patients with convulsions with gastroenteritis (CwG) or benign familial neonatal epilepsy (BFNE).43 ,48 ,50 BFNE is most commonly caused by mutations in the potassium channel subunit genes KCNQ2 and KCNQ3.61 The seizures in CwG show similar clinical characteristics to those seen in BFIE, but occur in the context of gastroenteritis, often caused by rotavirus infection.62
There is no evidence of a genotype–phenotype relationship between PRRT2 mutations and the three different phenotypes with which they are associated. All three phenotypes (BFIE, PKD, ICCA) are associated with the common insertion mutation (p.Arg217Profs*7) and all three phenotypes are also associated with other mutations, including the 15 missense mutations. The lack of a genotype–phenotype relationship is not unexpected given the phenotypic variability seen in families with ICCA, in which the same mutation can cause BFIE alone, PKD alone, or both syndromes in different individuals within the same family. This suggests that the expression of the phenotype is influenced by other genetic or environmental factors, rather than the particular PRRT2 mutation carried by an individual. What these additional factors may be has yet to be determined.
The common c.649-650insC mutation occurs in a homopolymer tract of nine cytosine bases that are preceded by four guanines. This sequence has the potential to form a hairpin loop (as illustrated in figure 3), which may cause polymerase slippage and the insertion of an additional base during DNA replication.25 The poly-C tract appears to be particularly prone to mutation: in addition to the common insertion mutation, a 1 bp deletion and a nonsense mutation affecting the same nucleotide have been reported several times (table 1). Furthermore, there are five single nucleotide polymorphisms (SNPs) altering bases within the poly-C tract reported in public databases (dbSNP and 1000 Genomes).
The high frequency of the c.649-650insC mutation in this homopolymer tract explains, at least in part, the failure of many previous efforts to identify the BFIE/ICCA/PKD gene. Insertions in homopolymer tracts are less likely to be detected by next generation sequencing (NGS) technologies, due to the technical limitations of the chemistries and detection methods used by them.63 Indeed, the failure of NGS methods to detect the common insertion mutation has been described twice.25 ,29 Homopolymer tracts can also affect the results of classical Sanger sequencing. Polymerase slippage when reading through homopolymers leads to ‘noisy’ or ambiguous sequence following the homopolymer tract. This has the potential to mask the presence of insertion mutations in the homopolymer or other mutations in the downstream sequence, as has been noted.27
Functional role of the PRRT2 protein
The full length PRRT2 protein contains 340 amino acid residues with two putative transmembrane domains near the C-terminal end (figure 2A). The expression of the gene has begun to be characterised.4 ,25 ,28 The mouse orthologue has been shown by in situ hybridisation analyses to be localised throughout the brain, with the highest mRNA concentrations seen in the cerebral cortex.4 ,25 Reverse transcriptase PCR (RT-PCR) experiments also showed Prrt2 expression in the brain, with lower values seen in spinal cord and no expression in other tissues tested.4 The Prrt2 mRNA concentrations in mice were highest on postnatal day 14 (P14), which corresponds approximately to an age of 1–2 years in humans. Western blots of mouse tissues probed with anti-PRRT2 antibody showed high expression in the brain, low expression in spinal cord, and no expression in other tissues tested,28 reiterating the RT-PCR results described above. Overall, these data demonstrate convincingly that PRRT2 codes for a protein with specific expression in the brain and nervous system. The expression of PRRT2 peaks during postnatal development, consistent with its role in the pathogenesis of infantile seizures. Robust expression of the mRNA throughout the mouse brain is still seen at postnatal day 46,25 approximately equivalent to adolescence in humans. The mRNA is present in the adult mouse brain, although the expression levels are approximately 50% of those seen at the peak of expression on P14.4 The expression of PRRT2 into adulthood is consistent with its role in the pathogenesis of PKD, which has onset in late childhood or adolescence and sometimes continues during adult life.
Although PRRT2 is largely an uncharacterised protein, the first steps have been taken in the understanding of its role in neuronal function. Yeast-2-hybrid studies suggested that it interacts with synaptosomal associated protein 25 (SNAP-25).64 This interaction was confirmed by in vitro and in vivo co-immunoprecipitation experiments.28 SNAP-25 is a member of the SNARE protein family. Proteins in this family are essential for the transport of vesicles through the Golgi apparatus and to the plasma membrane. SNAP-25 forms part of a complex involved in the release of neurotransmitters from synaptic vesicles at the presynaptic membrane. The SNAP-25 protein is located on the cytoplasmic surface of the plasma membrane in association with syntaxin-1, which is membrane bound. Neurotransmitter release is triggered by the influx of calcium ions resulting from an action potential. These activate synaptotagmin and synaptobrevin molecules on the cytoplasmic surface of the synaptic vesicle, causing the binding of synaptobrevin to syntaxin-1 and SNAP-25 and bringing the synaptic vesicle to the plasma membrane. The synaptic vesicle then fuses with the plasma membrane, releasing its contents into the synapse.65 Co-immunostaining experiments using FLAG tagged PRRT2 expressed in primary hippocampal neurones indicated that PRRT2 co-localised with synapsin-1 at neuronal puncta.28 Synapsin-1 associates with the cytoplasmic surface of synaptic vesicles and is involved in synaptogenesis and the modulation of neurotransmitter release.66 ,67 The localisation of PRRT2 at neuronal puncta and its interaction with SNAP-25 suggest that it may also play a role in the modulation of neurotransmitter release. A reduction in the amount of PRRT2 due to haploinsufficiency presumably leads to a dysregulation of this process. The seizures and dystonic posturing seen in BFIE and PKD could possibly result from either excessive neurotransmitter release at excitatory synapses or a reduction in the release of inhibitory neurotransmitters. Understanding which of these processes is affected will require a more precise understanding of the role of PRRT2 in synaptic transmission. Increased understanding of the pathogenic mechanism underlying PRRT2 mutations may also explain the particular effectiveness of some antiepileptic drugs—for example, carbamazepine, a sodium channel blocker—in treating the disorders associated with PRRT2 mutations. Presently there is no obvious link between decreased sodium channel activity and the effective treatment of a disorder resulting from altered neurotransmission, but this may become apparent in the future.
Mutations in PRRT2 account for between 40–100% of familial cases of BFIE,25 ,29 ,39 ,46 ,48 33–100% of familial cases of ICCA,22 ,24 ,28 ,30 ,34 ,35 ,39 ,46 ,48 and 62–100% of familial cases of PKD4 ,22–24 ,28 ,30 ,34 ,35 in the various cohorts of patients that have been studied. The percentages of sporadic cases found to be PRRT2 mutation positive are generally lower, with mutations found in 27–50% of PKD patients with no family history23 ,24 ,30 ,35 and 29–100% of cases of sporadic benign infantile seizures.25 ,29 ,39 ,46 ,48 The generally high frequency of mutations in familial cases of PKD and BFIE indicates that PRRT2 mutations are the most common cause of both disorders. The frequency of mutations is particularly high in ICCA. It is possible that the rare mutation negative ICCA families have non-coding mutations or large deletions affecting PRRT2, as these would not have been detected by the sequence based screening methods used for the studies reviewed here. It is therefore possible that all cases of ICCA are caused by PRRT2 mutations. The mutation negative BFIE and PKD cases may also have these types of mutation, or may have mutations in other genes.
PRRT2 is the major gene for BFIE, ICCA, and PKD and contains the second highest number of reported mutations associated with epilepsy after SCN1A. The vast majority (95%) of mutations in the gene are truncating mutations predicted to lead to haploinsufficiency. A small number of missense mutations have been reported and these all alter amino acid residues clustered in two predicted transmembrane domains at the C-terminal end of the protein. PRRT2 is predicted to code for a protein involved in the modulation of presynaptic neurotransmitter release, and a perturbation of this process is likely to be the cause of the seizure and movement disorder phenotypes associated with mutations in the gene. The identification of a heterozygous mutation in PRRT2 can provide a definitive diagnosis for patients with suspected BFIE, ICCA or PKD. This molecular diagnosis can reduce or prevent the need for additional investigations in these patients and guide treatment for these disorders.
The authors thank Associate Professor Paul Thomas for advice on the correlation of developmental ages in humans and mice.
Contributors Sarah Heron and Leanne Dibbens co-wrote this manuscript.
Funding This work was funded by National Health and Medical Research Council of Australia Training Fellowship 1016715 to SEH and Career Development Fellowship 1032603 to LMD.
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.