Febrile convulsions are a common form of childhood seizure. It is estimated that between 2 and 5% of children will have a febrile convulsion before the age of 5. It has long been recognized that there is a significant genetic component for susceptibility to this type of seizure. Wallace, Berkovic and co-workers recently reported linkage of a putative autosomal dominant febrile convulsion gene to chromosome 8q13–21. We report here another autosomal dominant febrile convulsion locus on chromosome 19p. Linkage analysis in this large multi-generational family gave a maximum pairwise lod score of 4.52 with marker Mfd120 at locus D19S177. Linkage to the chromosome 8 locus was excluded in this family. Haplotype analysis using both affected and unaffected family members indicates that this febrile convulsion gene, which we call FEB2, can be localized to an 11.7 cM, 1–2 Mb section of chromosome 19p13.3, between loci D19S591 and D19S395.
Childhood convulsions associated with febrile episodes are relatively common and represent the majority of childhood seizures. The National Institutes of Health (NIH) has defined a ‘febrile convulsion’ (FC) as a seizure event in infancy or childhood, usually occurring between 3 months and 5 years of age, associated with fever but without any evidence of intracranial infection or defined pathological or traumatic cause (1). Studies in the developed world indicate that between 2 and 5% of all children will experience an FC before the age of 5 (1,2). In certain Pacific populations the incidence rate can be as high as 15% (1,3).
Approximately 33% of FC patients will experience a second FC (4–8). Of this sub-population, 50% will go on to have a third. Only 9% of recurrent FC patients have more than three FCs (1,9). Between 2 and 7% of children who experience FCs go on to develop non-febrile seizure disorders and epilepsy later in life (9,10). This is 2-10 times higher than the general population. There is some controversy over whether FC patients go on to develop a higher incidence of mesial temporal sclerosis, the primary cause of temporal lobe epilepsy, and indications are that this progression may be linked to the duration of the FC (9).
It has long been known that FCs are more likely to occur with a higher frequency in families where there is already a documented history of FCs (1). Children from such families show a 3-fold or greater risk than the general population of experiencing FCs. The genetics of familial FCs is somewhat ambiguous and all inclusive at this time, with polygenic, autosomal dominant and autosomal recessive models all receiving some support (11–13). Most recently, a putative autosomal dominant gene for familial FCs was reported linked to chromosome 8q13-21 (14). This was the first report of a genetic locus being linked to FCs. We report here the absence of linkage to chromosome 8q13-21 in a large Midwestern family and the presence of an autosomal dominant febrile convulsion locus on human chromosome 19p.
Determination of seizure onset and frequency and phenotypic evaluation in the K4 family
The three generation familial febrile seizure kindred K4 is displayed in Figure 1. In this and all the families studied individual family members were considered affected if they demonstrated a positive history of having experienced one or more febrile convulsions before the age of 5. Positive history is defined primarily by parental recollection. Where possible, particularly in some of the older family members, the results were confirmed by interview with either siblings, the patient themselves or with the family physician where medical records were available.
In the family where we were able to subsequently demonstrate a significant lod score, the K4 family, the average age of onset for febrile seizures was between 6 months and 2 years of age, with the majority occurring during the first year of life. The exception was individual III:5, whose onset occurred after age 3. Average seizure duration was ∼15 min. The number of seizures ranged from a single seizure in one family member (III:5), between one and five seizures in four family members (III:10, III:11, IV:6 and IV:9), to over five seizures for five members of this family (II:5, III:2, III:3, IV:1 and IV:2). A detailed seizure history was unavailable for several, mostly older, members of this family. None of the affected family members in family K4 went on to develop any other form of epilepsy.
Exclusion of the putative C8q febrile convulsion locus
Analysis of the lod score for C8q markers in the putative FC locus defined by Wallace et al. (14 ) did not show any significant indication of linkage in any of the 19 febrile families screened (Table 1A). Our analysis yielded consistently negative lod scores for all three C8q markers screened. Using an autosomal dominant model in which susceptible genotypes had 95% penetrance we obtained a lod score of -6.15 ([theta] = 0.001) at D8S530 in the K4 family, the locus where Wallace and co-workers obtained their maximum lod score. Analysis of the same data using a 64% penetrance and a sporadic rate of 7% yielded results consistent with the original model.
Linkage mapping of C19p polymorphic markers
A sex equal linkage map of short tandem repeat polymorphisms (STRPs) for the region of interest from 19p is shown in Table 2. The map was based on typing short tandem repeat polymorphic markers through at least eight large CEPH families. This order was derived using the program CRIMAP. In each case pairwise marker order was established on the basis of at least one clear recombination event. Primer pairs D19S1034 (GATA21G05) and D19S594 (GATA66B01) amplify the same STRP.
DNA from 19 families containing multiple cases with FCs was analyzed using STRPs taken from Marshfield screening sets 4-7. After testing 130 polymorphic markers on all chromosomes a significant lod score was obtained with marker Mfd120 at locus D19S177 in the K4 family. Markers adjacent to D19S177 were also tested, again with only the K4 family, yielding maximum significant lod scores at θ = 0 (Table 1B). The lod scores obtained in the other febrile seizure families gave only negative or weakly positive scores at all [theta] values (Table 1B).
Additional STRPs mapping near Mfd120 on chromosome 19 were then targeted for screening in the K4 family. The maximum lod score was 4.52 at zero recombination with marker Mfd120 at locus D19S177 (Table 3). Maximum scores in excess of 3.0 at low [theta] values were obtained with three additional markers in this region, D19S534, D19S216 and D19S427 (Table 3). The majority of markers tested in the immediate region yielded positive maximum scores (lod ≥ 1.0) at low θ values (Table 3), consistent with linkage of a febrile convulsion susceptibility locus in this region.
Our choice of this genetic model is appropriate given the fact that families where the mode of inheritance was consistent with an autosomal dominant model were selected for participation in this study. However, in order to test the assumptions, additional analyses were performed under less stringent conditions (penetrance of at-risk genotypes reduced to 64%, all unaffected family members assumed to be diagnosis unknown). Results obtained under this model produced evidence consistent with that seen with the original model, but with the expected reduction in lod score (positive scores reduced, negative scores increased) and was still consistent with localization of a febrile seizure locus to this region of c19 (unpublished results).
Haplotype analysis and analysis of recombination events
Haplotypes in all the families studied were derived in individuals without regard to affectation status. The affected members of the K4 family had broad regions of shared haplotype (Fig. 1, black bars). Affected individuals or obligate carriers (assuming a dominant transmission model) II:1, II:3, II:5, III:2, III:3, III:5, III:9, III:11, IV:1, IV:2, IV:6 and IV:9 shared alleles at all loci listed in Table 1 and Figure 1. Individuals III:10, IV:3 and IV:10 were recombinant in the region of interest. Individuals III:10 and IV:10 were recombinant at D19S395. Patient III:10, an affected individual, had the ‘affected’ haplotype telomeric of D19S395 and ‘unaffected’ haplotype centromeric. Individual IV:10, an unaffected patient, had the ‘unaffected’ haplotype telomeric of D19S395 and the ‘affected’ haplotype centromeric. Patient IV:3, also unaffected, had the ‘affected’ haplotype telomeric of D19S247.
Individual IV:5 shared an extensive region of the haplotype with the other affected members of this family (Fig. 1). This individual's original diagnosis, as determined by the parent, was that he had an ‘event’ as a child but they were uncertain whether to label it a seizure. This patient was assigned a diagnosis of ‘status unknown’ for these analyses. Given the nature of the diagnosis and the frequently decades long interval between the febrile events and ascertainment of families for this study, the diagnostic data are in remarkable agreement with the haplotype analysis.
The common interval defined using crossovers in affected individuals extends from the telomere to D19S395, an interval of some 21.9 cM on the Marshfield c19 sex equal linkage map. Including the crossover seen in the unaffected individual IV:3 the smallest interval where the FEB2 gene is likely to be found extends from D19S591 to D19S395, spanning ∼11.7 cM on the 19p sex equal linkage map.
This current study shows significant evidence for linkage between the telomeric end of 19p and a gene for familial febrile convulsions. Linkage to the putative febrile convulsion locus at 8q13-21 was excluded in this family.
Three crossover events, one in affected member III:10 and two in unaffected individuals IV:3 and IV:10 (Fig. 1), helped to define the candidate region for FEB2 to an 11.7 cM segment of 19p13.3. On the Livermore c19 physical map this critical region spans only 1.5-2 Mb (15,16). This fairly large discrepancy in the linkage and physical distances can be explained by the relatively high meiotic recombination rate seen in this region of 19p (15,16; http://www-bio.llnl.gov/bbrp/genome/chrom_map.html).
Unlike the family described by Wallace et al. (14), there was a low incidence of subsequent epilepsy in the FC K4 kindred. Only one member of this pedigree, individual III:7, who has had no recorded febrile seizures, experienced a documented seizure later in life and this was clearly associated with severe head trauma. There does appear to be a high incidence of attention deficit-hyperactivity disorder in the family, but the lack of firm clinical diagnosis, particularly in the older family members, prevents us from drawing any further conclusions.
There are a number of potential candidate genes in this region of 19p13.3 and in the syntenic region of mouse chromosome 10. Among the candidates to be found in our region of interest are genes coding for the [alpha] subunits of the G binding protein (GNA11 and GNA15), the thromboxane A2 receptor (TBXA2R), zinc finger 57 (ZNF57) and the megakarocyte-associated tyrosine kinase gene (MATK). None of these can be considered intuitively obvious candidates for FEB2. In addition, there are at least two uncharacterized ESTs in this interval, EST01708 and EST02235 (15,16; http://www-bio.llnl.gov/bbrp/genome/chrom_map.html).
Two of the more intriguing candidates with a known neurological phenotype that fall within our region of interest are the jittery/hesitant locus on the homologous portion of mouse chromosome 10, syntenic to 19p13.3 (17), and the Caymen ataxia locus, an autosomal recessive cerebellar ataxia that shares a nearly identical critical interval with FEB2 (18). The clustering of interesting neurological genes to 19p13.3 makes this region a good candidate for detailed physical mapping and sequencing.
Patients from families with a history of FCs who in fact experience FCs as children themselves can go on to develop a higher incidence of generalized epilepsy than is seen in the general population. Such patients also have a higher incidence rate for idiopathic temporal lobe epilepsy associated with hippocampal sclerosis. The relatively common nature and potential long-term ramification of febrile seizures make understanding the familial form of the disease and potentially preventing FCs in this clinically significant sub-population an important biomedical priority. A more thorough understanding of the neuropharmacological/neurophysiological aspect of this disease will add to our knowledge of the fundamental mechanism involved in epileptogenesis as well.
Materials and Methods
Determination of phenotype
Families where the mode of inheritance was consistent with an autosomal dominant model were selected for participation in this study. Individual family members were considered affected if they demonstrated a positive history of having experienced one or more febrile convulsions before the age of 5. Positive history is defined primarily by parental recollection. Where possible, particularly in some of the older family members, the results were confirmed by interview with either siblings, the patient themselves or with the family physician where medical records were available.
Sample collection, seizure history and genotyping
A detailed Informed Consent Form approved by each participating institution's Institutional Review Board (IRB) was presented to all individuals participating in these studies. After obtaining informed consent two 10 ml blood samples were drawn by venipuncture from the family members.
In addition, each family member was requested to fill out a questionnaire detailing febrile seizure affectation status, number of seizures, age of onset, length of seizure, type of seizure, subsequent non-febrile seizure history and diagnosis, medication history and knowledge of other family members status
Genomic DNA was extracted from the lymphocytes as described elsewhere (19). Genotyping with STRPs was completed using methods previously reported (20). Markers were developed either at Généthon (21,22), at Marshfield, at the Jackson Laboratory (23) or within the Cooperative Human Linkage Center (24).
Linkage mapping, statistical analysis and haplotype analysis
Linkage maps were constructed using CEPH reference family genotyping data generated at Marshfield and data obtained from versions 6 and 7 of the CEPH databases, as well as unpublished information. Maps were built using the program CRIMAP (25). Pairwise lod scores were computed using LINKAGE version 5.03b (26). An autosomal dominant model of FC susceptibility was used, with the susceptible genotypes having 95% penetrance. An alternative model using 64% penetrance and assuming all unaffected individuals in the family were in fact of unknown diagnosis was also applied.
Haplotyping in all families was done without regard to individual affectation status. Analysis of haplotype was done both manually and using the haplotype subroutine of the Cyrillic software package (Cherwell Scientific).
The authors would like to thank the various members of our study family for their cooperation and enthusiastic participation. We gratefully acknowledge Dr Samuel Berkovic, Dr R.H.Wallace and their co-workers for allowing us access to a preprint of their C8q linkage paper several months before its publication. Finally, Mrs Mary Spindler here at the Marshfield Medical Research and Education Foundation was essential for collecting samples and isolating DNA from the K4 family presented here. This work was supported in part by NIH grant NS16308.
cM, centiMorgan; PCR, polymerase chain reaction; STRP, short tandem repeat polymorphism.