The first two authors contributed equally to this study.
Association Analysis of Malic Enzyme 2 Gene Polymorphisms with Idiopathic Generalized Epilepsy
Article first published online: 28 SEP 2005
Volume 46, Issue 10, pages 1637–1641, October 2005
How to Cite
Lenzen, K. P., Heils, A., Lorenz, S., Hempelmann, A. and Sander, T. (2005), Association Analysis of Malic Enzyme 2 Gene Polymorphisms with Idiopathic Generalized Epilepsy. Epilepsia, 46: 1637–1641. doi: 10.1111/j.1528-1167.2005.00270.x
- Issue published online: 28 SEP 2005
- Article first published online: 28 SEP 2005
- Accepted June 18, 2005.
- Idiopathic generalized epilepsy;
Summary: Purpose: Linkage disequilibrium mapping revealed allelic and haplotypic associations between single-nucleotide polymorphisms (SNPs) of the gene encoding the malic enzyme 2 (ME2) and adolescent-onset idiopathic generalized epilepsy (IGE). Homozygote carriers of the associated ME2 haplotype had a sixfold higher risk of IGE compared with any other genotype. The present population-based association study tested whether genetic variation of the ME2 gene confers susceptibility to common IGE syndromes in the German population.
Methods: The study included 666 German healthy control subjects and 660 German IGE patients (IGE group), of which 416 patients had an age at onset in adolescence (IGEado group). Genotyping was performed for six SNPs and one dinucleotide repeat polymorphism, all located in the ME2 region.
Results: Neither allele nor genotype frequencies of any ME2 polymorphism differed significantly between the controls and the IGE groups (p > 0.22). No hint of an association of the putative risk-conferring haplotype was seen, when present homozygously, in both IGE groups compared with controls (p > 0.18).
Conclusions: These results do not support previous evidence that genetic variation of the ME2 gene predisposes to common IGE syndromes. Thus if a recessively inherited ME2 mutation is present, then the size of the epileptogenic effect might be too small or not frequent enough to detect it in the present IGE sample.
Idiopathic generalized epilepsies (IGE) affect ∼0.3% of the general population and account for ∼30% of all epilepsies (1). The clinical features of IGE are characterized by an age-related occurrence of recurrent unprovoked generalized seizures in the absence of detectable brain lesions or metabolic abnormalities (2). The common IGE syndromes comprise childhood (CAE) and juvenile absence epilepsy (JAE), juvenile myoclonic epilepsy (JME), and epilepsy with generalized tonic–clonic seizures (EGTCSs) (2–4). The etiology of IGE is genetically determined, but the complex pattern of inheritance suggests an interaction of several susceptibility genes (5). Twin and family studies indicate an overlapping genetic component that is shared across the common IGE subtypes (6–8), but also provide evidence that specific gene configurations determine the particular IGE subtype (9–12).
A recent genome-wide linkage scan yielded strong evidence for a major locus common to most IGE syndromes at the marker D18S474 on chromosome 18q21.1 (maximum 2-point LOD score of 5.2, assuming a recessive mode of inheritance and evidence of locus heterogeneity) (10). Subsequent linkage disequilibrium mapping at the D18S474 locus revealed allelic and haplotypic associations between single-nucleotide polymorphisms (SNPs) of the gene encoding the malic enzyme 2 (ME2) and adolescent-onset IGE (13). The ME2 SNPs showed consistent evidence of strong allelic association, ranging from p values of 0.010 to 0.0001 in both case–control (156 IGE patients: 88 JME, 68 JAE or EGTCS, 126 randomly chosen controls) and family-based samples (59 with JME, 49 with JAE or EGTCS) of European origin. Moreover, homozygotes of a common ME2 haplotype, composed by nine SNPs across the ME2 locus, displayed a sixfold higher risk for IGE [odds ratio (OR) = 6.1; 95% confidence interval, 2.9–12.7). These findings suggest that ME2 is the susceptibility gene predisposing to IGE at the 18q21.1 locus. ME2 is a genome-coded mitochondrial enzyme that converts malate to pyruvate. Recessively inherited ME2 deficiency could indirectly decrease neuronal synthesis of the neurotransmitter γ-aminobutyric acid (GABA), in that it supplies pyruvate for GABA synthesis (14). Compelling lines of evidence show that an impaired GABAergic inhibition increases neuronal excitability and may contribute to epileptogenesis (15,16). The high frequency of the homozygous ME2-risk haplotype in the IGE group (35%) and the remarkable effect size (OR = 6) define the first common target for pharmacogenetic intervention (13). However, the underlying functional ME2 mutation remains to be identified. Considering the fact that none of the reported associations with IGE has been consistently replicated (17), the original association of ME2 polymorphisms and IGE requires further confirmation. The present population-based association study tested whether genetic variation of the ME2 gene predisposes to IGE in the German population.
The study participants were recruited at the Department of Neurology, Charité University Medicine in Berlin, and at the University Clinic of Epileptology in Bonn, Germany. The study protocol was approved by the Institutional Ethics Committees at each clinical site. Written informed consent was obtained from all participants, who were unrelated individuals of German descent. The diagnostic classification of IGE syndromes was performed according to the Classification of Epilepsies and Epileptic Syndromes, as proposed by the International League Against Epilepsy (2). The study sample included 660 IGE patients with the following syndromes: CAE (n = 201), JAE (n = 100), JME (n = 244), and epilepsy with generalized tonic–clonic seizures on awakening (EGMA, n = 115). Greenberg et al. (13) reported haplotypic association of ME2 polymorphisms with adolescent-onset IGE (JME, JAE, or EGTCS), but they did not show that this association was restricted to an age-related subgroup of IGE syndromes. Therefore we conducted association analyses in the entire IGE sample (n = 660) and in a subgroup with adolescent-onset IGE (IGEado, n = 416), comprising 212 JME, 100 JAE, and 104 EGMA patients. All 666 control subjects were screened for neuropsychiatric disorders by either a questionnaire or a standardized interview by an experienced neurologist/psychiatrist. Individuals with a history of epileptic seizures or major neuropsychiatric disorders were excluded from the control sample.
DNA samples of the study participants were isolated from whole blood or lymphoblastoid cell lines by using standard methods. The genotypes of six SNPs (SNP1–6) and one single tandem repeat polymorphism (STR, D18S46) were assessed for each study participant. The assignment and chromosomal location of the ME2 polymorphisms are shown in Table 1. SNP5 (rs649224) in exon 13 of the ME2 gene encodes a missense variation that alters amino acid at position 450 from glycine to glutamine (G450E). SNP3 and 5 also were part of the risk-conferring ME2-centered nine-SNP haplotype reported in the original association study (13). In addition, we selected SNP6 (rs2156010) from the original study to test a second association peak ∼70 kb telomeric to the ME2 gene (13). The STR D18S46, located in intron 4 of the ME2 gene, was chosen to increase ME2 haplotype diversity.
|Number||dbSNP||Variation||ME2 location||Position on chrom. 18 [kb]||Intragenic position [kb]|
Assays-on-Demand and Assays-by-Design products (Applied Biosystems, Foster City, CA, U.S.A.) were applied to assess SNP genotypes by using TaqMan nuclease assays (18). Allelic discrimination was carried out by measuring the fluorescence intensity of reporters at the polymerase chain reaction (PCR) end point, by using an ABI Prism 7900 HT System and the SDS software version 2.1 (Applied Biosystems).
The STR D18S46 was genotyped by standard polymerase chain reaction (PCR), by using one fluorescent-labeled primer, followed by capillary electrophoresis of the amplified fragments on ABI 3730 DNA Analyzers and allele-calling by the GENEMAPPER version 3.0 software (Applied Biosystems). The assignment of the D18S46 alleles refers to the allele set (GDB: 61853) reported in the GDB Human Genome Database (http://www.gdb.org/).
Allele and genotype frequencies, χ2 tests, and odds ratios (ORs) for individuals carrying the risk haplotype together with the 95% confidence interval (95% CI), and the test for Hardy–Weinberg Equilibrium were calculated by using the SAS computer program (19). Haplotype frequencies were estimated with the expectation-maximization algorithm implemented in the program COCAPHASE 2.35 (http://www.hgmp.mrc.ac.uk) (20), which was also used to evaluate haplotypic associations and to calculate pairwise linkage disequilibrium (LD) across the ME2 polymorphisms. Because the EM algorithm does not accurately estimate haplotype frequencies <1%, rare haplotypes in both groups were not considered (21). The D′ and r2 values were used to describe the magnitude of LD (22). A two-sided type I error rate of p = 0.05 was chosen for the analyses. Power analysis for a presumed frequency of the risk factor of 8% indicated that the sample sizes of both IGE groups and the controls provide a statistical likelihood of >93% to detect a susceptibility effect with an attributable relative risk of 2. A relative risk of 1.77 would be sufficient to detect an association with a power of 80% in the IGEado group.
The genotype and allele frequencies of the ME2 SNPs in the controls and IGE groups are shown in Table 2. The scoring rate of the genotyped individuals was >98% in the study samples. All duplicated samples (9.7%, internal genotyping controls) displayed the same genotypes as its reference sample for all ME2 polymorphisms, and consistent mendelian inheritance was demonstrated in eight control families. None of the genotype distributions in the controls deviated significantly from those expected by Hardy–Weinberg equilibrium (p > 0.45). Pairwise D′ values of 1 were observed between all intragenic ME2 polymorphisms, including the STR D18S46. SNPs 2, 3, and 4 were in complete LD to each other (D′= 1; r2=1). The common 129-bp allele of D18S46 was nearly in complete LD (r2= 0.99) with the major allele of SNPs 2, 3, and 4.
|SNP1||N||A/A||A/C||C/C||f(C)||P*, 1df||P*, 2df|
|SNP3||N||T/T||T/C||C/C||f(C)||P*, 1df||P*, 2df|
|SNP5||N||T/T||T/C||C/C||f(C)||P*, 1df||P*, 2df|
|SNP6||N||T/T||T/C||C/C||f(C)||P*, 1df||P*, 2df|
The allele and genotype frequencies of the ME2 SNPs did not differ significantly between the controls and the IGE groups (p > 0.22; Table 2). Likewise, no significant difference was found in the distribution of the D18S46 alleles between the controls and the IGE patients (χ2= 8.67; df= 10; p = 0.564) and the patients with adolescent-onset IGE (χ2 = 6.82; df= 9, p = 0.655). The genotype distributions of the ME2 SNPs were similar in the patients with adolescent-onset IGE and CAE patients (p > 0.33).
Haplotype analysis of the ME2 polymorphisms was carried out for SNPs 1, 3, and 5 because of the complete LD of SNPs 2, 3, and 4. SNPs 1, 3, and 5 distinguished three common ME2 haplotypes (Table 3). None of the estimated haplotype distributions differed significantly between the controls and each IGE group (p > 0.33; Table 3). Haplotyping of homozygotes is without ambiguity, because only one type of nucleotide occurs at each locus of the haplotype. We found no excess of homozygous three-SNP haplotypes in both IGE groups when compared with the controls (Table 4). In particular, the frequency of controls homozygous for the putative risk haplotype C_C_C was not significantly different from that observed in either the IGE patients (χ2= 1.738; df= 1; p = 0.187; OR, 1.17; 95% CI, 0.926–1.48) or the patients with adolescent-onset IGE (χ2= 0.835; df= 1; p = 0.361; OR, 1.13; 95% CI, 0.868–1.48). To increase haplotype diversity, haplotype analyses also were carried out with four ME2 polymorphisms by including the highly polymorphic dinucleotide repeat D18S46. D18S46 is a suitable marker to search for recessively acting susceptibility alleles, because of its heterozygosity rate of 77.7% and strong LD to ME2 SNPs that were part of the risk haplotype (13). Nonetheless, no evidence for an excess of homozygous D18S46 genotypes or homozygous four-marker haplotypes was found in the IGE groups compared with the controls (Table 4).
|Haplotype/Sample||N||A_T_C||A_C_T||C_C_C||P*, df= 2|
|Sample/Allele/ Haplotype||Controls (n = 658)||IGE (n = 645)||IGEado (n = 412)|
|D18S46 129 bp||0.138||0.149||0.149|
|D18S46 131 bp||—||0.003||0.002|
|D18S46 139 bp||—||0.002||—|
|D18S46 141 bp||0.005||0.003||0.002|
|D18S46 143 bp||0.008||0.012||0.012|
|D18S46 145 bp||0.061||0.074||0.066|
|D18S46 147 bp||0.012||0.006||0.002|
To evaluate a second association peak ∼70 kb telomeric to the ME2 gene (13), we conducted association analysis with SNP6 (rs2156010), which is located in the peak region. SNP6 displayed strong LD with SNP3 (D′= 0.80; r2= 0.48) and D18S46 (D′= 0.85) but a weak LD with SNP5 (D′= 0.13). No evidence for an allelic or genotypic association was found between SNP6 and both IGE groups (p > 0.70; Table 2).
The original haplotypic association implies that a common recessively inherited ME2 mutation confers a substantial susceptibility effect to common IGE syndromes (13). This finding is of great practical importance because the presumed epileptogenic ME2 genotype defines the first common pharmacogenetic target for therapeutic intervention. However, the original association has not been confirmed independently, and the underlying ME2 mutation remains to be identified.
The present association study failed to provide supportive evidence that genetic variation of the ME2 gene predisposes to common IGE syndromes. We found no hint of an allelic or genotypic association of the ME2 SNPs neither in the entire IGE sample (p > 0.22) nor in the adolescent-onset IGE group (p > 0.25), whereas the original study reported consistent evidence for strong allelic associations of all used SNPs in the ME2 region, ranging from p values of 0.010 to 0.001 (13). These controversial findings were also observed for SNP3 and 5, which were tested in both studies. Strong LD across the investigated SNPs, and the STR suggests that the ME2 gene is located within a single haplotype block, which displays low haplotype diversity. Two of six SNPs investigated in the present study were sufficient to distinguish each of the observed six-SNP haplotypes. The risk-conferring ME2-centered nine-SNP haplotype of the original study corresponds to the most common three-SNP haplotype (SNPs 1, 3, and 5: C_C_C) in the present study. The frequency of homozygous individuals carrying the risk-conferring haplotype differed marginally between the IGE patients in both studies (35% vs. 32–33%), whereas a remarkable difference was observed among the control subjects (8% vs. 29%). Considering that control subjects of both studies were of European origin, this finding implies that population stratification could lead to spurious association findings (17). When homozygotes carrying the C_C_C haplotype were considered, we found no hint for an association in the IGE groups compared with the controls (p > 0.18). Even if the nine-SNP ME2-centered haplotype is regarded as a subtype of the three-SNP haplotype C_C_C, we should not have missed a common recessively acting ME2 mutation with the reported effect size (OR = 6). To increase haplotype diversity, we also included the STR D18S46 into the haplotype association analyses. Although D18S46 is a highly informative dinucleotide repeat polymorphism, we observed strong LD with all ME2 SNPs (D′= 1) used in the present study. Thus it is reasonable to assume that one of the D18S46 alleles should also be in LD with the common ME2 mutation. With regard to the heterozygosity rate of 77%, D18S46 is a suitable marker to search for recessively acting disease alleles. Nonetheless, we found no excess of homozygous carriers of D18S46 genotypes in the IGE groups when compared with controls (Table 4). Similar frequencies of the genotype distribution in patients with either adolescent-onset IGE or CAE provided no evidence that the putative susceptibility allele confers an age-related effect to the epileptogenesis of common IGE syndromes.
Failure to replicate initial association findings in genetically complex traits does not necessarily disprove the existence of a susceptibility gene (see ref. 17 for a comprehensive discussion of controversial association findings). The first study often suggests a higher frequency and a stronger genetic effect of the underlying susceptibility allele than is found by subsequent studies (23). However, it is unlikely that we missed a substantial susceptibility effect because of lack of power. The present study samples were four times larger than those of the original study and should reliably detect a risk factor with an attributable relative risk of 1.77. Likewise, undetected population stratification appears to be unlikely to cause a false-negative association result in the present study samples of German origin. Recent SNP-based analysis (>140 SNPs) of genetic substructures in the German population gave no indication of population heterogeneity between two subsamples of 720 individuals with German-born grandparents collected from Southwest and Northwest regions of Germany (Krawczak et al., personal communication). To test for population stratification in the present case–control samples, we also applied a novel genomic control approach that assesses genomic similarities among unrelated individuals by using 26 unlinked, highly polymorphic STRs, which were chosen as ethnicity markers based on normative data from various European populations (24). Adaptive cluster analysis revealed well-matched individual distributions of allelic vector spaces among IGE patients and controls (data not shown). Thus even if cryptic substructures do exist in the present case–control samples, the effect of undetected population stratification seems to be negligible.
Taken together, we could not confirm a previous association, suggesting that a common recessively acting ME2 mutation predisposes to common IGE syndromes. Thus if a susceptibility effect of the ME2 gene exists, then the size of the effect might be too small or not frequent enough to detect it in the present IGE sample. Taking into account the potential pharmacogenetic implications, further replication studies will be necessary to verify the original association claim. Alternatively, mutation analysis of the risk haplotype could be a justifiable effort to identify the underlying ME2 mutation.
Acknowledgment: We are grateful to all study participants and all clinicians contributing to this study. We thank Carolin Engel for technical assistance. This work was supported by grants of the Deutsche Forschungsgemeinschaft (Sa434/3–1), the German National Genome Research Network (01GS0479, 01GS0474), the Michael Foundation, and the German Volkswagen Foundation.
- 7Syndrome-related genetics in generalized epilepsy. Epilepsy Res Suppl 1991;4: 105–11., .
- 19SAS Institute Inc. SAS/STAT User`s Guide, Release 6.03. Cary , NC : SAS, 1988.