|Home | About | Journals | Submit | Contact Us | Français|
Benign familial neonatal seizures are most often caused by mutations in the voltage‐gated potassium channel subunit gene KCNQ2. More than 60 mutations have been described in BFNS families, approximately half of which lead to protein truncation. The hypothesis of this study was that deletion or duplication of 1 exons of KCNQ2 could cause BFNS in cases without coding or splicing mutations.
Multiplex ligation‐dependent probe amplification (MLPA) was used to test a group of 21 unrelated patients with clinical features consistent with either BFNS, benign familial neonatal–infantile seizures or sporadic neonatal seizures, for exonic deletions and duplications.
Three deletions and one duplication mutation were identified in four familial cases and cascade testing of their available family members showed that the mutations segregated with the phenotype in each family. The junction fragment for one of the deletions was amplified by PCR and sequenced to characterise the breakpoint and verify that a deletion had occurred.
Submicroscopic deletions or duplications of KCNQ2 are seen in a significant proportion of BFNS families: four of nine (44%) cases previously testing negative for coding or splice site mutation by sequencing KCNQ2 and KCNQ3. MLPA is an efficient second‐tier testing strategy for KCNQ2 to identify pathogenic intragenic mutations not detectable by conventional DNA sequencing methods.
Benign familial neonatal seizures (BFNS; OMIM 121200) and benign familial neonatal–infantile seizures (BFNIS; OMIM 607745) are autosomal dominant seizure disorders of early infancy. In BFNS, affected neonates have onset of seizures in the first few days of life and offset usually occurs by a few months of age.1 BFNS is caused in most families by mutations in the voltage‐gated potassium channel subunit genes KCNQ2 (OMIM 602235) and KCNQ3 (OMIM 602232),2,3,4 and to date, >60 mutations in these genes have been reported in BFNS families. Most of these mutations are in KCNQ2.5,6 In contrast, the age of seizure onset in BFNIS varies from the neonatal period to 13 months of age.7 BFNIS, which is caused by mutations of the voltage‐gated sodium channel subunit gene SCN2A (OMIM 182390), is rarer than BFNS, with 11 families described to date.7,8,9,10 Although there is phenotypic overlap between these two disorders, they can be diagnosed on clinical grounds but with the probability of correct diagnosis increasing with the number of affected family members for whom the age of seizure onset is accurately known.
Approximately half of the KCNQ2 mutations described in BFNS are truncations, splice site defects, or deletions or insertions of a small number of bases causing frameshifts. All these are predicted to lead to haploinsufficiency, as no functional protein can be produced from the mutated allele. Larger deletions and duplications affecting 1 exons of KCNQ2 would also be predicted to cause BFNS. The chance finding of a submicroscopic deletion that removed an allele of a microsatellite marker closely linked to KCNQ2, causing apparent aberrant segregation, was originally exploited to identify KCNQ2 as a BFNS gene.4 This deletion was later found to remove the last nine exons of the KCNQ2 gene and 22 kb of the 3′ genomic sequence.11 However, no other mutations altering the copy number of 1 contiguous exons have subsequently been described because this type of mutation is not detectable using standard, systematic PCR‐based mutation screening and sequencing methods.
The development of multiplex ligation‐dependent probe amplification (MLPA) allows the relatively rapid and inexpensive screening of patients for deletions, duplications and other copy number variants affecting any number of contiguous exons in a selected gene by testing for changes in the copy number of each exon.12 We used this technology to screen for intragenic copy number variations in KCNQ2 in a group of unrelated patients in whom KCNQ2 and KCNQ3 mutations had been previously excluded by direct sequencing.
In total, 21 probands were tested, of whom 18 had neonatal seizures. Of these 18 probands, 8 had classic BFNS, 7 had sporadic neonatal seizures and 3 had familial neonatal seizures with atypical features such as refractory ongoing epilepsy and developmental delay. The patients with neonatal seizures had previously been tested by direct sequencing for mutations in all coding exons of KCNQ2 and exons 5 and 6 of KCNQ3. Three probands were from families diagnosed with BFNIS and these patients had also been sequenced for all coding exons of SCN2A. The initial clinical diagnosis of BFNS versus BFNIS in families with both neonatal and early infantile onset (less than 6 months) was determined as follows. Where the majority of cases had a neonatal onset and a single case or small number of cases had early infantile onset, BFNS was diagnosed. Where early infantile and neonatal onset were in similar proportions, or early infantile predominated, a clinical diagnosis of BFNIS was made. DNA was extracted from peripheral blood samples using standard methods. The study was approved by the ethics committee of Austin Health.
Samples were prepared for MLPA by repurifying 1 µg of DNA using the Qiaquick PCR purification kit (Qiagen, Valencia, California, USA). MLPA was carried out using the SALSA MLPA kit P166 KCNQ2 (MRC Holland, Amsterdam, The Netherlands) according to the manufacturer's instructions. This kit contains probes specific for 14 of 17 exons and the promoter of KCNQ2 (probes for exons 2, 8 and 14 are not included) and 13 control probes specific for other regions of the human genome. Automated analysis, using GeneMarker V.1.5 software (SoftGenetics, State College, Pennsylvania, USA) comparing peak heights was carried out on files directly imported from an ABI 3100 Avant DNA analyser (Applied Biosystems, Foster City, California, USA).
Long‐range PCR to search for amplifiable junction fragments was carried out using Platinum Taq High Fidelity DNA Polymerase (InvitrogenCarlsbad, California, USA). PCRs were set up and run according to the manufacturer's instructions. Reactions contained between 40 ng and 100 ng of template DNA. The reactions were cycled for 35 cycles with an extension time of 4 minutes and a 5‐minute final extension following cycling. Products were separated by electrophoresis on 0.8% agarose gels and visualised by ethidium bromide staining. Products for sequencing were purified using the Qiaquick PCR purification kit (Qiagen). Sequencing reactions were carried out using BigDye V. 3.1 (Applied Biosystems) and analysed on an ABI 3730 capillary sequencer (Applied Biosystems).
Three deletions and a duplication were identified in 4 of the 21 patients tested (fig 11).). Of the 14 unsolved familial cases, deletions or duplication mutations involving multiple exons occurred in 3 classic BFNS families and 1 initially suspected on clinical grounds of being BFNIS. No mutations were present in the remaining seven sporadic cases of neonatal seizures.
The three deletions detected all removed multiple contiguous exons of KCNQ2: in family A, exons 3–11 were deleted; in family B, exons 13–15; and in family D exons 1–5. The duplication in family C involved an extra copy of exons 3–12, inserted in unknown orientation (fig 22).). All available relatives of these patients were tested by MLPA. The respective deletions were detected in nine affected members in family A and all four affected members and one obligate carrier (D‐III‐2) in family D. The deletion in family B was detected in both affected members but was not seen in either of the unaffected members of the first generation of this family and therefore represents a de novo mutation. Segregation of 12 highly polymorphic microsatellite markers was consistent with the pedigree as given. Four affected members of family C were tested, and the duplication was present in all of them. The pedigrees of the four families are shown in figure 33.
Of the four families with mutations, two (B and C) were typical BFNS families on clinical grounds. Family D was initially ascertained with one confirmed neonatal onset and one definite infantile onset case and was tentatively diagnosed as BFNIS; subsequently, two more relatives with neonatal onset were ascertained. This, and the molecular result showing the deletion of KCNQ2 exons 1–5, led to the definitive diagnosis as BFNS. Family A has eight members who had neonatal seizures with onset of seizures in the first month of life. Seizures were focal clonic or generalised tonic–clonic with apnoea and all ceased by 12 months of age. An additional three family members (A:II:4, A:III:4 and A:III:5) had seizures commencing in adulthood. Two members had focal seizures that secondarily generalised, while the third had several nocturnal tonic–clonic seizures that could not be classified. None of these three members experienced seizures in the neonatal period (although the mother of A:II:4 was not available for questioning). The clinical details of the affected members of the four families are summarised in table 11.
As there were no prior samples with KCNQ2 deletions available to validate the MLPA procedure, we chose one case to independently confirm the result. LR‐PCR was carried out to characterise the breakpoints of the smallest deletion, which was in family B. The primers 5′‐GCTTCCTCTCTGAAGTGTGTCAG‐3′ and 5′‐TACTGCAGGAGAGACTCTCAACAG‐3′ were used to amplify a product spanning both deletion breakpoints for the deletion that excised exons 13–15. The expected normal PCR product from these primers is 3849 bp in length. This product was not successfully amplified; however, a smaller product of approximately 800 bp in length was amplified from the homologue affected by the deletion from two affected members of family B (fig 4A4A).). This confirmed the presence of the deletion detected by MLPA and indicated that the size of the deletion in family B was approximately 3 kb. The product amplified from individual B:II:2 was sequenced to determine the exact excision points for the deletion, the exact size of the deletion and if additional nucleotides were incorporated during the DNA repair that rejoined the sequence (fig 4B4B).). The deletion in family B was IVS12‐768 to IVS15+572 del3018bp, effectively validating the MLPA result. No additional nucleotides were added to the sequence during DNA repair. The exact size of the PCR product was 831 bp.
The occurrence of submicroscopic deletions and duplications at the intragenic level of resolution was predicted for KCNQ2, based on the high proportion of known truncating mutations in the gene. MLPA was used to screen a group of families tentatively diagnosed on clinical grounds as BFNS or BFNIS, and a group of patients with sporadic neonatal seizures, in whom KCNQ2 mutations had not previously been identified by PCR‐based techniques. Submicroscopic deletions and a duplication mutation were detected in three of eight likely BFNS families and one of three possible BFNIS families, definitively diagnosing the latter as BFNS based on the molecular result and bringing the total number of BFNS families in the study to nine. The same mutations were detected in all affected relatives of each proband. No mutations were detected in the sporadic cases with neonatal seizures. The detection rate for deletions and duplications in patients with BFNS who had previously tested negative by sequencing for KCNQ2 and KCNQ3 mutations was 44%.
For one mutation, the deletion breakpoints were characterised. The DNA repair in this case was not accompanied by nucleotide insertions. A 4 bp nucleotide insertion was previously characterised as part of the DNA repair process associated with an SCN1A deletion detected by MLPA.13 Few such breakpoints have been characterised.
All four mutations detected are very likely to be pathogenic, causing the loss of one functional copy of the KCNQ2 gene, and therefore haploinsufficiency. This mechanism has been suggested for BFNS2 and is the likely result of the numerous truncating mutations of KCNQ2 identified in BFNS families. The KCNQ2 protein forms part of the inhibitory M‐channel and therefore reduction in its activity, resulting from either missense mutations or haploinsufficiency, is expected to increase neuronal excitability and therefore seizure susceptibility.
Two families (C and D) each contained a single person who had onset of seizures at 4 months of age. BFNS families with KCNQ2 mutations have been reported in which occasional members had seizure onset between 1 and 8 months of age.11 Family C has an extensive family history of seizures, all commencing in the neonatal period, therefore the single individual with later seizure onset did not modify our diagnosis of BFNS (see Methods), but was considered to fall within this previously observed variation. However, family D is much smaller, initially with only two confirmed affected members. One had neonatal onset and one had infantile onset, leading to the initial diagnosis of BFNIS. This illustrates that the clinical distinction between BFNS and BFNIS is difficult in small families with both neonatal and infantile onsets. A family with a KCNQ2 mutation has previously been reported in which the onset of seizures occurred between 2 and 4 months of age, consistent with a familial diagnosis of BFNIS.14 Together with our data, this further highlights the phenotypic overlap between families with KCNQ2 and SCN2A mutations and the need to incorporate genetic testing as part of the patient investigation if a definitive clinicomolecular syndromal diagnosis is required. Detection of the KCNQ2 defect in family D and the ascertainment of two more affected members with neonatal seizure onset suggested that the diagnosis of BFNS for the family is more likely.
Family A has two members who did not have observed neonatal seizures, confirmed by a reliable maternal history, but did have focal seizures commencing in adulthood. Their mother also had seizures in adult life, but her neonatal history was unknown. These three people (A:II:4, A:III:4 and A:III:5) all have the BFNS‐related KCNQ2 intragenic deletion. The mutation may be responsible, at least in part, for the partial epilepsy in these members as an alternative phenotype.
Other than the families included in this study, we have examined 15 BFNS families in whom missense or truncating mutations were previously detected by sequencing (Biervert et al2, Richards et al15 and unpublished data). Submicroscopic deletions or duplications were detected in four of the nine BFNS families that were studied here by MLPA, bringing the total number of BFNS families for whom we have identified mutations to 19 out of a total of 24 studied. These numbers suggest that submicroscopic deletions and duplications in KCNQ2 account for a significant proportion of unsolved familial cases in which mutations are not detected by direct sequencing: over one‐fifth of all KCNQ2 mutations and one‐sixth of BFNS families.
A small proportion of probable BFNS families remains unsolved. These could be caused by pathogenic mutations in the promoter or introns of KCNQ2, or additional genetic heterogeneity. The promoter and most of the intronic sequences are not usually covered by PCR‐based mutation screens, but mutations in these regions could lead to downregulation of transcription or incorrect splicing of the KCNQ2 transcript. There may also be deletions or promoter mutations in KCNQ3, although mutations leading to haploinsufficiency are less likely in KCNQ3 given that all the mutations described to date in this gene are missense mutations.5 Another possibility is the presence of at least one other BFNS gene, the existence of which is likely to be confirmed only if a family is identified that is large enough for exclusion of linkage to both KCNQ2 and KCNQ3 or by demonstration of linkage to another region of the genome. A final possibility is that a proportion of cases could be polygenic, which might include large copy number variations affecting a number of susceptibility genes related to channel function or expression, including KCNQ2 and KCNQ3.
These results suggest that MLPA is an important second‐tier testing strategy for suspected cases of BFNS or BFNIS without previously identified KCNQ2,KCNQ3 or SCN2A coding or splice‐site mutations. Submicroscopic deletions and duplications of KCNQ2 may account for a significant proportion of these cases.
We thank the patients and their families for their participation in this study, Tristiana Williams and Bev Johns for technical assistance and Dr Dan Casalaz, Mercy Hospital for Women for referral of families. This work was supported by grants from the National Health and Medical Research Council of Australia.
BFNIS - benign familial neonatal‐infantile seizures
BFNS - benign familial neonatal seizures
MLPA - multiplex ligation‐dependent probe amplification
OMIM - Online Mendelian Inheritance in Man
Competing interests: none declared.