Analysis of Mutations in 7 Genes Associated with Neuronal Excitability and Synaptic Transmission in a Cohort of Children with Non-Syndromic Infantile Epileptic Encephalopathy

Epileptic Encephalopathy (EE) is a heterogeneous condition in which cognitive, sensory and/or motor functions deteriorate as a consequence of epileptic activity, which consists of frequent seizures and/or major interictal paroxysmal activity. There are various causes of EE and they may occur at any age in early childhood. Genetic mutations have been identified to contribute to an increasing number of children with early onset EE which had been previously considered as cryptogenic. We identified 26 patients with Infantile Epileptic Encephalopathy (IEE) of unknown etiology despite extensive workup and without any specific epilepsy syndromic phenotypes. We performed genetic analysis on a panel of 7 genes (ARX, CDKL5, KCNQ2, PCDH19, SCN1A, SCN2A, STXBP1) and identified 10 point mutations [ARX (1), CDKL5 (3), KCNQ2 (2), PCDH19 (1), SCN1A (1), STXBP1 (2)] as well as one microdeletion involving both SCN1A and SCN2A. The high rate (42%) of mutations suggested that genetic testing of this IEE panel of genes is recommended for cryptogenic IEE with no etiology identified. These 7 genes are associated with channelopathies or synaptic transmission and we recommend early genetic testing if possible to guide the treatment strategy.


Introduction
Epileptic Encephalopathy (EE) is a heterogeneous condition in which cognitive, sensory and/or motor functions deteriorate as a consequence of epileptic activity, which consists of frequent seizures and/or major interictal paroxysmal activity [1]. This concept was formally recognized in 2001 and subsequent International League Against Epilepsy (ILAE) reports. The 2010 report of the ILAE Commission on Classification and Terminology stated that "Epileptic Encephalopathy patients with definite evidence of brain insult, confirmed disorder of cortical development by magnetic resonance imaging, neurocutaneous disorders, syndromal disorders and confirmed or highly suspected neurometabolic disorders based on clinical (multi-system involvement including organomegaly or skeletal changes) and biochemical markers. Extensive neurometabolic evaluations conducted for all these patients were negative (blood for amino acid, biotinidase, ammonia, lactate, glucose, very long chain fatty acids including phytanic and pristanic acids, transferrin isoform electrophoresis, total homocysteine, copper, coeruloplasmin, creatine and guanadinoacetate; urine for purine and pyrimidine screening, creatine, guanadinoacetate and organic acid; cerebrospinal fluid for glucose, lactate, protein, amino acid, neurotransmitters and 5-methyltetrahydrofolate). All patients failed to show any positive response to a trial of intravenous pyridoxine up to 300 mg under electroencephalography monitoring, followed by adequate trials of oral pyridoxine, pyridoxal phosphate and folinic acid [19]. We also excluded patients who fit into distinct electroclinical syndromes proposed by the ILAE and those not actively followed up in our centre.
Data variables collected from the medical charts included demographic information (gender, ethnicity, age at seizure onset and latest follow up), family history (febrile convulsion, epilepsy, intellectual disability and other neurological diseases), epilepsy details (seizure types at onset and latest follow up, seizure frequency and evolution, history of status epilepticus, antiepileptic medications used), neurological examination findings (upper motor neuron syndrome, hypotonia, movement disorders [dystonia, choreoathetosis, myoclonus, ataxia, parkinsonism], microcephaly, macrocephaly, dysmorphism), investigations (MRI brain and EEG results), mortality and other associated clinical features (autism spectrum disorder and other neurobehavioral disorders such as attention deficit hyperactivity disorder, visual impairment, hearing impairment, ability of independent walking and oromotor dysfunction requiring nasogastric tube or gastrostomy feeding). Information regarding the developmental status at the time of seizure onset and latest follow up was collected as well. Either formal neuropsychological testing (using Griffiths Mental Developmental Scale or HK-WISC) or best clinical assessment (based on developmental milestones recorded in the medical charts) were used to classify development or intelligence as normal, mildly delayed, moderately delayed or severely delayed.
All patients were screened for mutations of 6 genes (ARX, CDKL5, KCNQ2, SCN1A, SCN2A and STXBP1). Mutation analysis of PCDH19 was only performed in female patients as the PCDH19-associated X-linked IEE mainly affect female with heterozygous mutations.

Point mutation analysis
Genomic DNA samples of the patients were extracted from peripheral blood using Flexigene DNA Kit (Qiagen GmbH, Germany). All exons covering the coding regions as well as the splice junctions were amplified by polymerase chain reaction (PCR) using oligonucleotide primers designed based on the reference genomic sequence (Table 1) of different genes. PCR contained 0.1 μg of genomic DNA as template, 5 pmol of each primer, 200 μM of deoxyribonucleoside triphosphates, and 0.5 U HotStarTaq Plus DNA polymerase (Qiagen) in 1X Qiagen PCR buffer. PCR was carried out with initial enzyme activation at 95°C for 5 minutes, followed by 50 cycles of denaturation at 94°C for 30 seconds, annealing at 60°C for 1 minute and extension at 72°C for 1.5 minutes, with a final extension at 72°C for 10 minutes. For those templates with high degree of secondary structures or high GC-contents, 1x Q-Solution (Qiagen) was included in the PCR mixtures. If the non-specific products could not be eliminated by adding Q-Solution, a higher initial activating temperature of 98°C and denaturing temperature of 96°C were used ( Table 1). The quality and quantity of PCR products were checked by agarose gel electrophoresis. PCR products were directly used for sequencing reaction by Bigdye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystem, Foster City, CA) and analyzed on a 3730xl DNA analyzer (Applied Biosystems).
Homology analyses with the reference genomic sequence (Table 1) were performed using NCBI program BLAST. The numbering for each mutation was taken from the start codon with +1 corresponding to the A of the ATG in the reference sequence (Table 1). Mutations were discriminated from single nucleotide polymorphisms (SNP) with allele frequency > 0.01 reported in NCBI SNP and Ensembl SNP database. The parental DNA was collected and sequenced to distinguish between de novo and familial variants.

Pathogenicity assessment of the mutations
Evolutionary conservation analysis was performed to predict whether the amino acid substitution in missense mutations would affect protein function based on the degree of conservation at the affected residues. Besides, online sequence homology-based tool, Sorting Intolerant from Tolerant (SIFT), Polymorphism Phenotyping v2 (PolyPhen-2) and Align-Grantham variation/ Grantham deviation (Align-GVGD) analysis have been used to predict whether the mutation would interfere with the protein function. We have described them previously [20]. The two splice site mutations in intron regions were analyzed by another online software tool, the Automated Splice Site Analyses (Laboratory of Human Molecular Genetics and Genomic Disorders, UWO, CA, https://splice.uwo.ca/) [21].

Multiplex ligation-dependent probe amplification (MLPA)
For identification of copy number variations (CNVs) of the intragenic regions or entire genes, MLPA was used. It made use of a single PCR primer pair for all the probes to determine the copy number of all sequences in a single multiplex PCR based reaction followed by capillary electrophoresis.
For the ARX, CDKL5, KCNQ2, PCDH19 and SCN1A genes, commercial MLPA probemixes (SALSA P189 CDKL5 probemix for both CDKL5 and ARX, SALSA P166 KCNQ2 probemix for KCNQ2 and SALSA P137 SCN1A probemix for SCN1A, MRC Holland, Amsterdam, The Netherlands) as well as the SALSA MLPA reagent kit (MRC Holland) were used. For SCN2A and STXBP1, commercial MLPA kits were not available and synthetic MLPA probes were designed by the online software H-MAPD suggested by the protocol (http://bioinform.arcan.stonybrook. edu/mlpa2/cgi-bin/mlpa.cgi) and synthesized for all the exons according to the guidelines and protocol provided by MRC Holland. The MLPA procedures were performed according to the manufacturer's protocols. Fragment analysis of PCR products was performed on the ABI 3130xl capillary sequencer (Thermo Fisher Scientific, Waltham, MA) by using GeneScan TM-500LIZ as size standards (Thermo Fisher Scientific) and HiDi formamide (Thermo Fisher Scientific). The GeneScan results were analyzed using Coffalyser software (MRC Holland). The peak area of a given exon was divided by the sum of 12 reference peak areas for each individual sample. The final ratio was obtained by dividing this relative peak area of the corresponding exon by the averaged normal control peak area. Thresholds of <0.65 were set for deletions and >1.35 for duplications.

Results
Clinical characteristics (Table 2) Twenty-six IEE patients, whose etiology was unknown and who did not fit into specific electroclinical syndromes, were identified from our registry. The majority (24/26, 92%) of patients were from Asia. Only 2 patients had an ethnic origin other than Asian. One was African-French (patient 32). Another was Portuguese-Chinese (patient 44). The majority (21/26, 81%) of the selected IEE cohort were pure Chinese. The mean age of seizure onset was 4.1 months (SD 3.9 months), while the median was 3 months. Epileptic spasm was the most common type of seizure at onset. 10 out of 26 (38%) had epileptic spasm as the first seizure type. Most of the children (92%) developed multiple types of seizures with time. Concerning the seizure evolution, 16 patients (62%) had more than 50% seizure reduction and 5 patients (18%) had 25-50% seizure reduction. Concerning neurodevelopment, 19 patients (73%) were severely developmentally delayed at the latest evaluation. Seven patients (27%) had developmental regression, which was considered to be the hallmark of EE and was defined as loss of acquired skills. This was compatible with EE which has the tendency to abate, discontinue or even stop, but often with serious neurocognitive deficits [22]. It was worth noting that 10 patients (38%) had a movement disorder exclusively in the form of dystonia, either generalized or focal and 7 patients (27%) had an upper motor neuron syndrome.

Mutation analysis
Eleven out of 26 patients (42%) were found to have mutations among the 7 genes (Fig 1 and  Table 2). All mutation details have been summarized in Table 3. Most of the variants were identified to be truncating and only one of them (p.A40V) was a missense mutation. Evolutionary conservation analysis showed that the affected amino acid residue of this missense mutation was highly conserved (Fig 1). SIFT, Polyphen-2 and Align-GVGD (Grantham variation: 0; Grantham deviation: 65.28; Class 65) analyses predicted that the missense mutation is probably damaging. Nine of the mutations were novel and the 2 CDKL5 mutations have been reported previously [23][24][25][26]. The three splice site mutations, IVS24-1G>T, IVS9-2A>G and IVS6 +1G>C, were predicted to form a leaky acceptor splice site, abolish the acceptor or donor splice site respectively. There were no microdeletions or CNVs for ARX, CDKL5, KCNQ2, PCDH19 and STXBP1 by MLPA analysis.

Discussion
High occurrence of mutations in our selected IEE gene panel We had attempted to apply the panel approach of genetic testing on subjects with IEE. These seven genes were reported to be involved in IEE. We found that the yield of the current gene panel analysis on IEE patients with unknown etiology was up to 42%. This is also the first  Mutation in SCN1A, SCN2A & KCNQ2 for neuronal excitability Ion channelopathies play a prominent role in the development of IEE. To date, more than 600 variants of SCN1A encoding for a voltage-gated sodium channel have been identified and most mutations were found in patients with DS [27]. Our previous study identified more than 70% of SCN1A mutations in a group of Chinese children with DS [20]. In contrast to SCN1A, much fewer SCN2A mutations were reported in previous literatures. Mutations were recently identified in severe EE including DS, infantile spasm and Ohtahara syndrome [28][29][30][31]. Our group also identified a SCN2A mutation in a patient with infantile spasm and severe intellectual disability previously [32]. In the present study, one SCN1A mutation and one microdeletion involving both SCN1A and SCN2A were identified. They were deleterious to the protein and the abnormal sodium channel function leading to severe phenotypes could result from haploinsufficiency as suggested previously [33,34]. Mutations of KCNQ2 encoding the voltage-gated potassium channels were identified in patients with neonatal EE [8]. KCNQ2 is expressed in broad regions of the brain and the gene products form heteromultimeric channels that mediate the M-current that inhibit the neuronal excitability [35]. Two IEE patients in the present study were found to have KCNQ2 mutations. The first mutation was a deletion-insertion mutation that replaced a short fragment LRPYD by two amino acids (PT) in the protein. The short fragment (LRPYD) is located in a highly conserved domain (A-domain) of C-terminal of KCNQ2 necessary for subunit interactions to form homo-or heteromeric channels to reach the surface [36,37]. Another KCNQ2 mutation was a splice site mutation which may cause aberrant splicing and disrupt the protein at the position within transmembrane domain 6.
Mutation in PCDH19 & STXBP1 for synaptic transmission PCDH19 belongs to the PCDHδ2 subgroup of PCDH family consisting of 6 extracellular cadherin (EC) repeats. It is involved in calcium-dependent cell-cell adhesion at the synaptic membrane [38,39] and it was hypothesized that the cellular interference was the main pathogenic mechanism associated with PCDH19 mutations [40]. Previously, we have identified PCDH19 mutations in two of our patients [20]. In the present study, a PCDH19 frameshift mutation (p. N846fsX861) was identified. This mutation terminates the protein at the cytoplasmic domain and abolishes the conserved CM1 and CM2 motifs [41,42]. Wolverton & Lalande [42] suggested that CM2 may play a functional role for mediating intracellular signal transduction. STXBP1 encoding for the neural-specific syntaxin-binding protein has long been discovered for regulation docking and fusion of synaptic vesicles through interaction with syntaxin in the SNARE complex for neurotransmitter release [43,44]. Until recently, STXBP1 mutations were identified to be associated with different forms of early-onset EE including Ohtahara syndrome, West syndrome and infantile spasms [7,[13][14][15]. STXBP1 is a horse-shoe shaped protein with 3 domains while domain 1 and 3a form the central cavity providing the binding surface for syntaxin [45]. In the present study, c.79delG is a novel frameshift mutation forming a stop codon in the early reading frame and IVS9-2A>G is a novel splice site mutation that may possibly disrupt the protein at domain 3a necessary for syntaxin binding.

Role of CDKL5 and ARX mutation in synaptic development
The CDKL5 protein belongs to the family of serine/threonine kinases which is characterized by an N-terminal catalytic domain [46]. In the past ten years, CDKL5 mutations were found to be associated with early-onset EE. In the present study, a relatively high percentage of CDKL5 mutations (14%) was found in non-syndromic IEE patients. p.A40V found in the present study is one of the mutation hot-spots located at the highly conserved ATP-binding site (amino acid  of the catalytic domain reported previously in 4 different studies including 5 patients [23][24][25][26]. Although mutation hot-spots were found in the catalytic domain, many pathological alterations can still be found in the C-terminal region [46]. The frameshift (p.K776fsX799) and nonsense (p.Q832X) mutations identified in the present study may cause truncation of the Cterminus. The 2 truncating mutations located upstream from 2 and 3 putative sites which are essential for the cellular localization of the protein [47]. Evidence was provided previously that the C-terminus of CDKL5 is a negative regulator of catalytic activity of CDKL5 and required for a proper subnuclear localization by protein-protein interactions [47][48][49]. p.K776fsX799 has been reported previously and immunofluorescence data of the same study demonstrated that the truncated protein mislocalized to the cytoplasm [23]. The important role of CDKL5 for proper brain function and development elucidate the relationship of CDKL5 mutations with neurodevelopmental disorders.
ARX encodes an important transcription factor that plays a significant role in the neuronal development of the brain [50]. In the present study, a heterozygous ARX mutation has been found in a female IEE patient with multiple seizure types, spastic dystonic quadriplegia and severe developmental delay. Although most affected females with ARX mutations showed relatively mild clinical outcomes as compared to males, severe cases were reported previously with various outcomes [51,52]. These cases may have occurred due to skewed X-inactivation or post-zygotic mosaicism [52]. Further studies will be performed to illustrate the pattern of X-inactivation in the patient. The previous literature reported ARX mutation associated with IEE in female only rarely and only 2 cases with truncating ARX mutations have been reported previously [51,53].

Association of the seven IEE genes with synaptic transmission
In the present study, mutations were found in the genes involved in neuronal excitability (KCNQ2, SCN1A, and SCN2A), synaptic transmission (PCDH19, STXBP1) and synapse development (ARX, CDKL5). The study of relationship between neurotransmitter release and ion channels illustrated that impairment in structure and function of ion channels can actually modulate the synaptic transmission by changing the synaptic terminal excitability [54,55]. The genetic defects found in the 7 genes may contribute directly or indirectly to the malfunction of synaptic transmission that may be an important mechanism for IEE. A recent comprehensive exome-sequencing study suggested that dysregulation of synaptic transmission plays an important role in the pathogenesis of EE as they demonstrated a significant enrichment of de novo mutations in genes annotated to be involved in synaptic transmission by pathway analysis [56].

Recommended diagnostic flow for patients with IEE
Based on the findings in our study, we propose a diagnostic algorithm for patients with IEE. Through clinical history taking, physical examination and neuroimaging (magnetic resonance imaging of the brain), relatively straightforward etiologies can be identified. If an underlying etiology is still unknown, a detailed neurometabolic evaluation should be performed especially aiming for potentially treatable causes such as vitamin-responsive epilepsies. This should involve an adequate trial (dosage and duration) of pyridoxine, pyridoxal phosphate and folinic acid. For those patients still without an underlying cause found, molecular workup is recommended. Candidate gene(s) testing can be performed according to the recommendation by Ottman et al [57] if a patient fits into a certain electroclinical syndrome. Otherwise, depending on the availability of resources, mutation analysis of our selected panel of genes (ARX, CDKL5, KCNQ2, PCDH19, SCN1A, SCN2A and STXBP1) is an option which can have a yield of up to 42%. Sanger sequencing of the selected gene panel is a relatively simple and direct method which do not require various steps of library preparation and target capturing, platforms for next generation sequencing, bioinformatics and various filtering strategies. Besides, the problem of uneven coverage is one of the issues that have to be overcome in next generation sequencing. Sanger sequencing of the selected gene panel will be a good choice for small-scale mutational studies with fewer resources available. However, for the remaining patients without any positive yield, next generation sequencing is still the choice for identification of other causative genes.
In the present study, except for the finding of the whole gene deletion of SCN1A and SCN2A in one patient, all of the MLPA analysis showed negative results for the other 5 genes. Besides, negative MLPA results have been found for other putative IEE-associated genes including NRXN1, GRIN2A and GRIN2B in our study (in preparation). As the yield of genetic defects identified by MLPA was low, we do not suggest trying to identify copy number variations of IEE-associated genes by MLPA if the resources are limited.

Conclusion
This study highlights that patients with non-syndromal IEE might not have specific phenotypes to guide candidate gene(s) selection. The yield of mutation analysis of seven selected genes of the IEE panel in this group of patients was 42%. Panel approach of genetic testing can be useful in investigating the underlying cause of IEE that do not fit into any distinct electroclinical syndromes and without any obvious etiologies including neurometabolic diseases.