Two-stage genome-wide association study identifies variants in CAMSAP1L1 as susceptibility loci for epilepsy in Chinese.

Abstract

In the majority of patients, epilepsy is a complex disorder with multiple susceptibility genes interacting with environmental factors. However, we understand little about its genetic risks. Here, we report the first genome-wide association study (GWAS) to identify common susceptibility variants of epilepsy in Chinese. This two-stage GWAS included a total of 1087 patients and 3444 matched controls. In the combined analysis of the two stages, the strongest signals were observed with two highly correlated variants, rs2292096 [G] [P= 1.0 × 10(-8), odds ratio (OR) = 0.63] and rs6660197 [T] (P= 9.9 × 10(-7), OR = 0.69), with the former reaching genome-wide significance, on 1q32.1 in the CAMSAP1L1 gene, which encodes a cytoskeletal protein. We also refined a previously reported association with rs9390754 (P= 1.7 × 10(-5)) on 6q21 in the GRIK2 gene, which encodes a glutamate receptor, and identified several other loci in genes involved in neurotransmission or neuronal networking that warrant further investigation. Our results suggest that common genetic variants may increase the susceptibility to epilepsy in Chinese.

Full text

Page 1

Two-stage genome-wide association study
identifies variants in CAMSAP1L1 as susceptibility
loci for epilepsy in Chinese
Youling Guo1, Larry W. Baum4, Pak Chung Sham1, Virginia Wong2, Ping Wing Ng6,
Colin Hiu Tung Lui7, Ngai Chuen Sin8, Tak Hong Tsoi9, Clara S.M. Tang1, Johnny S.H. Kwan1,
Benjamin H.K. Yip1, Su-Mei Xiao3, G. Neil Thomas10, Yu Lung Lau2, Wanling Yang2,
Stacey S. Cherny1,∗and Patrick Kwan5,∗
1Department of Psychiatry and The State Key Laboratory of Brain and Cognitive Sciences,2Department of Paediatric
and Adolescent Medicine and3Department of Medicine and Research Centre of Heart, Brain, Hormone & Healthy
Aging, The University of Hong Kong, Hong Kong, China,4School of Pharmacy and5Division of Neurology, Department
of Medicine and Therapeutics, Prince of Wales Hospital, The Chinese University of Hong Kong, Hong Kong, China,
6Department of Medicine and Geriatrics, United Christian Hospital, Hong Kong, China,7Department of Medicine,
Tseung Kwan O Hospital, Hong Kong, China,8Hospital Authority Head Office, Hong Kong, China,9Department of
Medicine, Pamela Youde Nethersole Eastern Hospital, Hong Kong, China and10Public Health, Epidemiology and
Biostatistics, University of Birmingham, Birmingham, UK
Received May 18, 2011; Revised November 11, 2011; Accepted November 18, 2011
In the majority of patients, epilepsy is a complex disorder with multiple susceptibility genes interacting with
environmental factors. However, we understand little about its genetic risks. Here, we report the first genome-
wide association study (GWAS) to identify common susceptibility variants of epilepsy in Chinese. This two-
stage GWAS included a total of 1087 patients and 3444 matched controls. In the combined analysis of the two
stages, the strongest signals were observed with two highly correlated variants, rs2292096 [G] [P 5 1.0 3
1028, odds ratio (OR) 5 0.63] and rs6660197 [T] (P 5 9.9 3 1027, OR 5 0.69), with the former reaching
genome-wide significance, on 1q32.1 in the CAMSAP1L1 gene, which encodes a cytoskeletal protein. We
also refined a previously reported association with rs9390754 (P 5 1.7 3 1025) on 6q21 in the GRIK2 gene,
which encodes a glutamate receptor, and identified several other loci in genes involved in neurotransmission
or neuronal networking that warrant further investigation. Our results suggest that common genetic variants
may increase the susceptibility to epilepsy in Chinese.
INTRODUCTION
Affecting up to 1% of people, epilepsy is the most common
serious chronic neurological disorder. Twin studies suggest
that epilepsy is highly heritable (1). Although a number of
familial epilepsy syndromes are recognized to result from
single-gene mutations, in the majority of patients, epilepsy is
thought to be a complex disorder with multiple susceptibility
genes interacting with various environmental factors that
include acquired CNS insults or underlying structural brain
abnormalities (e.g. stroke, head trauma, tumor) (2). Despite
drug treatment, up to 30% of patients have persistent seizures
(3). Discovery of genetic variants predisposing to the
development of epilepsy would advance our understanding
of epileptogenesis, leading to new drug targets, and facilitate
the evaluation of potentially anti-epileptogenic therapies
by targeting genetically susceptible individuals following
CNS insults.
∗To whom correspondence should be addressed at: Department of Psychiatry, The University of Hong Kong, Hong Kong, China. Tel: +852 28199581;
Fax: +852 28199550; Email: cherny@hku.hk (S.C.); Department of Medicine and Therapeutics, Prince of Wales Hospital, Hong Kong, China.
Tel: +852 26322211; Fax: +852 26321546; Email: patrickkwan@cuhk.edu.hk (P.K.).
# The Author 2011. Published by Oxford University Press. All rights reserved.
For Permissions, please email: journals.permissions@oup.com
Human Molecular Genetics, 2011
doi:10.1093/hmg/ddr550
1–6

HMG Advance Access published December 9, 2011
at Head Serials and Electronic Resources Dept on January 6, 2012
http://hmg.oxfordjournals.org/
Downloaded from

Page 2

There have been many attempts to identify the genetic sus-
ceptibility variants for the common epilepsy syndromes using
association studies of candidate genes selected either for their
role in monogenic epilepsy syndromes or on limited under-
standing of the pathobiology that underlies epileptogenicity
or seizure propagation. Disappointing results of these studies
(4) argue for a genome-wide approach without a priori
assumptions, which may discover previously unsuspected
markers. So far only one genome-wide association study
(GWAS) of epilepsy has been reported, which was conducted
in European subjects with partial (focal) epilepsy of both
known (‘symptomatic’) and unknown (‘cryptogenic’) causes
(5). Though the quantile–quantile (Q–Q) plots showed a
slight departure from normal expectation, none of the
P-values in their study reached the genome-wide significance
threshold. Because of differences in genetic structures between
different ethnic populations, it is possible that some genetic
factors influencing susceptibility to epilepsy may also differ.
Here, we report the first GWAS of epilepsy in Chinese.
RESULTS
In the discovery stage, testing for population stratification
using EIGENSOFTand principal components analysis
(PCA) found 15 significant principal components (P , 0.05)
which explained the largest proportion of inter-individual vari-
ation. These were controlled for in the subsequent genome-
wide associationanalysis.
Figure S1 shows the plot of the first two eigenvectors, indicat-
ing ancestry difference and individual admixture among the
study sample. A Q–Q plot showed an excess of P-values
expected under the null hypothesis (genomic inflation factor
l ¼ 1.15), which was reduced (l ¼ 1.02) after adjustment
for populationstratification
Fig. S2A). The Manhattan plot showed several genomic
regions as potential risk loci (Supplementary Material,
Fig. S2B), although none was genome-wide significant.
In the replication stage, among the 80 single-nucleotide
polymorphisms (SNPs) analyzed, 7 were nominally significant
with one-tailed P , 0.05 (Table 1 and Supplementary Mater-
ial, Table S3), and one was significantly associated with
epilepsy after Bonferroni correction (rs2292096 in 1q32.1).
In the combined meta-analysis of the discovery and replication
sample sets, this SNP in CAMSAP1L1 achieved genome-wide
significance [rs2292096 [G], P ¼ 1.0 × 1028, odds ratio
(OR) ¼ 0.63], and a second, highly correlated SNP (r2¼
0.87) was nearly significant (rs6660197 [T], P ¼ 9.9 × 1027,
OR ¼ 0.69) (Table 1). The OR for the two SNPs in the two
stages showed no significant heterogeneity (Phet. 0.1).
We used SNP imputation to investigate association strength
at untyped markers near the significant loci. Considering
CAMSAP1L1, there was a block of SNPs in tight linkage dis-
equilibrium (LD) (r2. 0.8) with much stronger significance
than any other SNPs at this locus, encompassing a 124 kb
region likely to contain the causative variant(s) responsible
for the association observed (Fig. 1A). Moderate association
wasobservedfor threegenotyped
P ¼ 6.8 × 1026; rs12742404, P ¼ 6.8 × 1026; rs6660197,
SupplementaryMaterial,
(SupplementaryMaterial,
SNPs(rs2292096,
Table 1. Details of SNPs with one genome-wide significant association and candidate loci with moderate associations in the discovery and replication cohorts, listed by chromosome and position order
Chr
SNP
Position
Gene
Minor
allele
Discovery cohort
Replication cohort
Combined
cohorts
Case
MAF
Control
MAF
OR (95% CI)
P-value
Case
MAF
Control
MAF
OR (95% CI)
P-value∗P-value
1q32.1a
rs6660197
198 970164 CAMSAP1L1 T
0.139
0.200
0.65 (0.53–0.79) 4.83 × 10250.161
0.203
0.75 (0.60–0.94) 0.00599
9.89 × 1027
1q32.1a
rs2292096
199 093392 CAMSAP1L1 G
0.120
0.184
0.61 (0.49–0.75) 6.76 × 10260.139
0.194
0.67 (0.53–0.85) 0.00038
1.04 × 1028
2p12b
rs4853352
77993 057
SNAR-H
G
0.246
0.185
1.48 (1.25–1.74) 1.54 × 10260.193
0.177
1.12 (0.89–1.39) 0.169
1.88 × 1025
2p12b
rs2164851
78155 723
SNAR-H
C
0.181
0.125
1.48 (1.23–1.79) 2.44 × 10250.258
0.207
1.16 (1.00–1.34) 0.026
5.27 × 1025
2p11.2b
rs2367575
86574 352
KDM3A
G
0.076
0.042
1.68 (1.27–2.22) 5.41 × 10250.054
0.037
1.50 (0.97–2.31) 0.033
4.93 × 1025
2q34b
rs13021324 212 058490 ERBB4
C
0.273
0.331
0.73 (0.63–0.86) 0.000162
0.311
0.346
0.86 (0.72–1.03) 0.048
5.8 × 1025
5p13.2b
rs6872795
35713 142
SPEF2
A
0.157
0.119
1.47 (1.21–1.79) 0.00013
0.121
0.093
1.33 (1.01–1.75) 0.022
1.81 × 1025
6q21c
rs9390754
102 071607 GRIK2
G
0.396
0.330
1.39 (1.20–1.60) 1.71 × 10250.323
0.332
0.96 (0.80–1.15) 0.668
0.0015
6q21c
rs4840200
102 433996 GRIK2
T
0.434
0.504
0.73 (0.64–0.85) 4.08 × 10250.474
0.469
1.02 (0.86–1.21) 0.592
0.0017
7q31.31brs702416
119 942506 KCND2
C
0.430
0.497
0.76 (0.66–0.87) 0.00012
0.482
0.510
0.89 (0.75–1.06) 0.095
0.00013
21q22.2brs1980406
40665 509
DSCAM
C
0.068
0.036
1.78 (1.32–2.40) 1.8 × 1025
0.049
0.031
1.61 (1.02–2.54) 0.021
1.66 × 1025
Position is as Genome Build 36.3 in base pairs. Chr, chromosome band; CI, confidence interval. Three types of associations are presented:aCAMSAP1L1, with SNPs reaching genome-wide significance in the
combined analysis;bsix loci with moderate associations observed, which involve genes that represent candidate genes of epilepsy;ca previously reported association, GRIK2, which is refined with SNPs
attaining significant P-values among prime candidate genes in the discovery cohort.
∗One-tailed P-value in the replication stage.
2Human Molecular Genetics, 2011
at Head Serials and Electronic Resources Dept on January 6, 2012
http://hmg.oxfordjournals.org/
Downloaded from

Page 3

P ¼ 4.8 × 1025) and six imputed SNPs (P , 3.5 × 1026) at
the locus in the discovery sample set (Fig. 1A).
We also examined the P-values from the discovery stage
obtained for 1710 SNPs of 194 prime candidate genes previ-
ously investigated for possible association with epilepsy (4)
(Q–Q plot shown in Supplementary Material, Figure S4).
Details of the SNPs and genes examined are provided in Sup-
plementary Material, Table S4. Three SNPs in GRIK2 on 6q21
were significantly associated, and another nearly so, after con-
trolling the false discovery rate (FDR) for 1710 tests, with
rs9390754 attaining P ¼ 1.7 × 1025; rs7747072, P ¼ 3.6 ×
1025; rs4840200, P ¼ 4.1 × 1025; and rs9390790, P ¼
1.2 × 1024. Association testing of imputed SNPs revealed
two regions with P-values close to those of the above four gen-
otyped SNPs within GRIK2. Two GRIK2 SNPs (rs9390754 and
rs4840200) were genotyped in the replication phase but
revealed no significant difference between the minor allele
frequencies in cases and controls (Supplementary Material,
Table S3). The 6q21 locus between rs9390754 and rs4840200
is marked by three recombination hot spots, which makes it pos-
sible that there are combined effects of the two SNPs (Fig. 1B).
The two SNPs are weakly correlated (r2¼ 0.001, D′¼ 0.034)
and are separated by several LD blocks. Multiple logistic
regression analysis showed that rs9390754 (P ¼ 0.0001) and
rs4840200 (P ¼ 0.0002) had independent effects. A 2 df likeli-
hood ratio test that accounted for the genotypic additive effects
of the two SNPs combined also showed strong evidence of the
combined effects (P ¼ 3.8 × 1027).
Associations with additional SNPs in six loci were
replicated with moderate P-values, suggesting that other risk
variants with a modest effect remain to be identified. These
may include variants located on chromosomes 2p12, 2p11.2,
2q34, 5p13.2, 7q31.31 and 21q22.2, which did not reach
genome-wide significance but could be considered suggestive-
ly associated with epilepsy (Table 1). In the 2p12 locus, eight
SNPs lying in a 224 kb region near SNAR-H achieved moder-
ate levels of association in the discovery set (P , 1024;
Supplementary Material, Fig. S3A). The strongest association
in this region was observed with the genotyped marker
rs4853352 in the discovery stage (P ¼ 1.5 × 1026). One
marker (rs2164851) in the SNAR-H gene also achieved
nominal significance in the replication study (P ¼ 0.026) and
showed a moderate level of significance in the combined ana-
lysis (P ¼ 5.3 × 1025). The regional association plots of these
suggestive regions are represented in Supplementary Material,
Figure S3A.
Figure 1. Association and LD plots of regions located on (A) 1q32.1 in a 250 kb interval in the CAMSAP1L1 gene, showing a genome-wide significant asso-
ciation, and on (B) 6q21in a 672 kb interval on chromosome 6, refining a previous association signal in the gene GRIK2. Discovery cohort association signifi-
cance in the regions is plotted against the left-hand y-axis as –log 10 (P-value). Genotyped SNPs were tested using the Cochran–Armitage trend test; imputed
SNPs were tested using a regression analysis based on the imputed allelic dosage; both tests were adjusted for population clusters. Genetic coordinates are as
NCBI Genome Build 36.3. The diamond denotes the association hit; rectangles, genotyped SNPs; circles, imputed SNPs; color scale, LD with the hit; purple line
and right-hand y-axis, recombination rate (cM/Mb as per HapMap data in CHB population). LD plots for the highlighted regions are based on HapMap CHB
release 22 using the pair-wise correlation coloring scheme of Haploview.
Human Molecular Genetics, 2011 3
at Head Serials and Electronic Resources Dept on January 6, 2012
http://hmg.oxfordjournals.org/
Downloaded from

Page 4

DISCUSSION
Studies attempting to identify susceptibility genes have tended
to focus on the idiopathic epilepsy syndromes, which usually
occur in children/adolescents and are generally drug respon-
sive. However, little attention has been paid to symptomatic
epilepsy, which is more often drug resistant, constituting a
major unsolved public health burden (3). It has been proposed
that causation of epilepsies can be regarded as a biological
continuum, and the degree of genetic contribution may differ
among different syndromes, being greater in idiopathic than
symptomatic syndromes (6). Genetic predisposition to symp-
tomatic epilepsy is not an entirely novel concept, although
very few studies have attempted to identify the genetic
markers associated with increased risk of epilepsy following
CNS insults. This hypothesis is supported by twin studies
showing pairwise concordance for symptomatic generalized
epilepsy and for partial epilepsy (7). APOE 14 allele carriers
were found to have a 2.4-fold increase in risk of epilepsy fol-
lowing traumatic brain injury (8).
In this two-stage GWAS of epilepsy in Chinese, combined
analysis showed the strongest signals with two highly corre-
lated variants, namely rs2292096 and rs6660197 on 1q32.1
in the CAMSAP1L1 gene, with the former reaching genome-
wide significance. Imputation revealed a block of SNPs
likely containing the causative variant(s), although none of
the SNPs examined is in the translated regions of the gene.
CAMSAP1L1 (CAMSAP1-like 1; also called CAMSAP1L2
or KIAA1078) is a calmodulin-regulated spectrin-associated
protein (CAMSAP) that belongs to a novel family of cytoskel-
etal proteins of little-known function. CAMSAP1 has been
reported to be expressed in neurons and astrocytes in the mam-
malian nervous system, where it is suggested to interact with
intermediate filaments (9). Further work has identified a
unique structural domain common to the three members of
the protein family that is able to inhibit neurite extension,
most likely by blocking microtubule function (10).
Using results of the discovery stage, we also refined a pre-
viously reported association with rs9390754 in the GRIK2
gene on 6q21. GRIK2, also called GluR6, is the glutamate re-
ceptor 6 gene, one of the high-ranking candidate genes for epi-
lepsy. Knockout mice deficient in the kainate-selective GRIK2
subunit of the kainate receptor had reduced susceptibility to
kainate-induced seizures (11). Alteration in GRIK2 mRNA
editing in neocortical tissue reflects an adaptive reaction to
ongoing seizure activity and may play a role in pathological
processes which contribute to seizure maintenance (12).
Forced overexpression of the GRIK2 kainate receptor within
the hippocampus induced seizures (13).
Weaker signals were observed in other loci, some of
which involved genes that could also represent candidate
genes of epilepsy (Table 1). These include KCND2, which
encodes the voltage-gated potassium channel Kv4.2, a key
component of the A-type potassium currents in the CNS
that critically regulate action potential back propagation
and the induction of specific forms of synaptic plasticity
(14). Kv channels are increasingly recognized to play an
important role in the pathogenesis of epilepsy. Specifically,
Kv4.2 knockout mice demonstrated increased susceptibility
to seizure induction (15). Dynamic alterations in Kv4.2
channel expression and localization were observed in a
variety of focal lesions associated with refractory epilepsy
in humans (16). Another potential locus was found in
ERBB4, whichisrecognized
susceptibility gene but has also been reported to be
mutated in a case of early myoclonic encephalopathy (17).
It encodes ErbB4, which is a member of the type I receptor
tyrosine kinase subfamily that promotes synapse formation
of GABA-containing interneurons in the hippocampus (18).
Other suggestive signals were found in genes previously
unsuspected but which could plausibly be associated with
epilepsy based on limited knowledge of their functions.
Theyinclude
SNAR-H,a
NF90-associated RNA family expressed in many human
tissues, with minor distribution in the brain (19). LRRTM4,
?400 kb from SNAR-H, is abundant in the dentate gyrus and
helps regulate cell–cell contact (20), which could plausibly
be involved in neuronal connectivity in epilepsy. DSCAM,
on 21q22.2, is suggested to play an important role in neural
circuitry development by allowing neurite ‘self avoidance’,
which refers to the tendency of branches from the same
neuron to selectively avoid one another (21). DSCAM knock-
out mice have dysregulated central respiratory function
because of impaired neural synchroneity; whether the abnor-
mality extends to other parts of the brain is unclear (22).
KDM3A (also called JMJD1A or JHDM2A) on 2p11.2 is a
histone demethylase that may reprogram neural stem cells
(23). SPEF2 (also known as KPL2) is preferentially expressed
in tissues that contain axonemal structures such as the brain,
lung and testis, and as with CAMSAP1L1, has a calponin hom-
ology domain (24).
The present study was limited by a small number of cases,
so that only associations with relatively large effect could be
detected. Given that the associations found in the present
study were not detected (Supplementary Material, Table S2)
in a previous GWAS of focal epilepsy in patients of European
ancestry (5), it is possible that they are unique to the Chinese
population. Allele frequencies in controls in Europeans (0.118
for both CAMSAP1L1 SNPs; see Supplementary Material,
Table S2) are much lower than those in Chinese controls
(0.200 and 0.184) and even lower than in Chinese cases
(0.139 and 0.120). Still, the frequencies in European cases
are slightly lower, in the same direction of effect as found in
the present Chinese sample. The discrepant results between
the two GWASs might also be due to differences in epilepsy
phenotypes. The European study, employing a heterogeneous
sample, likely only investigated genetic factors shared across
partial epilepsies, disregarding the type of epilepsy (idiopathic,
cryptogenic or symptomatic). For instance, 27% of patients in
the European study had hippocampal sclerosis and the cause of
epilepsy was unknown in 41% (5). In comparison, the most
common structural-metabolic cause in the present study was
stroke (17% of patients), and only patients with symptomatic
epilepsy were included in the discovery stage, although the
replication stage also included patients with cryptogenic epi-
lepsy, which may have a different pathogenesis. Our Hong
Kong cases appear to be recruited from a more homogeneous
and narrowly defined group of patients, and it is possible that
syndrome-specific common genetic causes do exist and were
detected in the present study.
asa schizophrenia-
member ofthesmall
4 Human Molecular Genetics, 2011
at Head Serials and Electronic Resources Dept on January 6, 2012
http://hmg.oxfordjournals.org/
Downloaded from

Page 5

In conclusion, this GWAS identified common genetic var-
iants that may increase the susceptibility to epilepsy in
Chinese. CAMSAP1L1 and the other genes where suggestive
loci were found might be considered candidate genes
because of their known or potential role in neurotransmission,
neuronal networking and connectivity. These findings lend
support to the concept that epileptogenesis may result from
a distributed hyperexcitable circuitry rather than a homogen-
ous epileptogenic focus (25). We suggest that a GWAS in
larger cohorts of Chinese subjects should be performed to
confirm the findings, and studies should be conducted to
explore the mechanisms for these associations.
MATERIALS AND METHODS
Study participants
Epilepsy patients of Han Chinese ethnicity aged between 2 and
91 years were recruited from neurology clinics of five regional
hospitals in Hong Kong covering a combined catchment popu-
lation of approximately 3 million. Exclusion criteria included
significant psychiatric comorbidity, history of pseudoseizures,
alcohol or illicit drug abuse, and presence of progressive or de-
generative neurological or systemic disorders. Syndromic clas-
sification wasadaptedfrom
organization of phenotypes in epilepsy (26). The study
included a total of 1087 Chinese epilepsy patients and 3444
ethnically matched controls. The discovery stage included
504 patients with symptomatic focal epilepsy and 2947 ethnic-
ally matched controls. All epilepsy patients were recruited in
Hong Kong. Non-epilepsy controls were recruited from two
sources: subjects recruited for other studies conducted in the
University of Hong Kong, genotyped with the same
platform (n ¼ 1947), and healthy individuals recruited in
Taiwan (n ¼ 1000). All subjects in the replication stage
were recruited in Hong Kong. They consisted of 583 patients
with either symptomatic or cryptogenic focal epilepsy and 497
controls who were healthy blood donors kindly contributed by
the Hong Kong Red Cross. Supplementary Material, Table S1
provides the clinical characteristics of the cases included in the
data analysis after quality control filters. The study was
approved by ethics committees of the participating hospitals,
and all patients or their legal guardians gave written informed
consent.
therevisedinternational
Genotyping and quality control
Genotyping of the discovery cohorts was performed using the
Illumina platforms at deCODE Genetics, Iceland (http://www.
decode.com). Cases and Hong Kong controls were genotyped
with the HumanHap 610-Quad BeadChip, and Taiwan
controls were genotyped using the HumanHap 550-Duo
BeadChip. The results were then merged using PLINK
(http://pngu.mgh.harvard.edu/purcell/plink). Common SNPs
typed in both groups were identified by filtering against the
HumanHap 550K Quad chip. A total of 88 subjects (16
cases, 72 controls) were excluded according to the following
criteria: (i) genotyping call rate ,95% (n ¼ 7); (ii) very
strong positive or negative autosomal heterozygosity (n ¼
15); (iii) related or duplicated individuals (n ¼ 25); (iv) sex
discrepancies (n ¼ 21); and (v) outliers in a plot of multidi-
mensional scaling analysis (n ¼ 20). Nearly half a million
common autosomal SNPs (n ¼ 461 024) passed the quality
control thresholds of .95% call rate, .1% minor allele fre-
quency (MAF) and Hardy–Weinberg equilibrium (HWE, P
≥ 0.0001). The total genotyping call rate in included indivi-
duals was 99.83% for cases and 99.85% for controls.
In the replication stage, SNPs with the lowest P-values from
the discovery stage were selected as follows. Genotyping
assays were designed for a Sequenom MassARRAY iPlex
System at the Hong Kong University Genome Research
Centre, Hong Kong (http://genome.hku.hk). SNPs ,100 kb
from other selected SNPs with lower P-values were not
initially selected. SNPs that could not be pooled together for
genotyping within three pools were replaced with other
SNPs within 100 kb or, if none, other SNPs with the next
lowest P-values. The process was repeated until no more
SNPs could be grouped into three pools, at which point
there were 82 SNPs. These were genotyped. After quality
control measures, 29 subjects (12 cases, 17 controls) were
excluded owing to low call rates. In addition, results of two
SNPs were excluded, one due to a call rate ,95% and one
due to violation of HWE (P , 0.0001) in the controls, result-
ing in 80 SNPs for analysis.
Statistical analysis
In the discovery stage, EIGENSOFT (http://helix.nih.gov/App
lications/eigensoft.html) (27) and PCA were used to control
for population stratification. For the correction of population
structure, we excluded a subset of SNPs (n ¼ 363 904) in ap-
proximate LD with each other (r2. 0.2) before running PCA,
and then obtained correlation matrices among remaining
SNPs. Genome-wide association analysis was performed for
thefullsetofSNPs,usingancestry-adjustedgenotypesandphe-
notypes,controllingforthe15significantcomponents.Associa-
tions with P , 5 × 1028
significant, as is generally accepted. Q–Q plots were con-
structed by contrasting uncorrected and corrected experimental
P-value distributions to the expected uniform 0–1 distribution.
Inthereplicationstage,case–controlanalysiswasperformed
using the Cochran–Armitage trend test as implemented in
PLINK (28). Meta-analysis of the discovery and replication
cohortswasthenperformedusinganinverse-variance-weighted
method under a fixed-effects model as implemented in PLINK
(28). Homogeneity for SNP effect across the studies was
tested using the Cochran Q test (29).
Imputation analyses were performed with IMPUTE v2
(https://mathgen.stats.ox.ac.uk/impute/impute_v2.html)
taking data from CHB + CHD individuals from HapMap 3
as the reference set of haplotypes. We analyzed only regions
surrounding significantly or marginally associated SNPs that
were either genotyped or could be imputed with relatively
high calling confidence (.90%). Association analysis of
imputed SNPs was performed assuming an underlying additive
model and including the first 15 EIGENSOFT eigenvectors as
covariates, which accounted for uncertainty in prediction of
the imputed data by weighing genotypes by their posterior
probabilities.
were considered genome-wide
(30),
Human Molecular Genetics, 20115
at Head Serials and Electronic Resources Dept on January 6, 2012
http://hmg.oxfordjournals.org/
Downloaded from

Page 6

The combined analysis of the two-stage study had .80%
power to identify SNPs conferring genotypic relative risks of
1.5–4.5 with minor allele frequencies of 0.01–0.5 at a ¼
5 × 1028. Power calculations were performed using CaTS
(www.sph.umich.edu/csg/abecasis/CaTS/).
SUPPLEMENTARY MATERIAL
Supplementary Material is available at HMG online.
ACKNOWLEDGEMENTS
We are grateful to all participants who contributed to this
study. We also thank Professor Annie W.C. Kung, Department
of Medicine, The University of Hong Kong, for contributing to
the control data in the discovery stage.
Conflict of Interest statement. The authors declare no compet-
ing financial interests.
FUNDING
The study was supported by the Research Grants Council of
the Hong Kong Special Administrative Region, China
(project numbersHKU7623/08M,
CUHK4466/06M).
HKU7747/07Mand
REFERENCES
1. Kjeldsen, M.J., Kyvik, K.O., Christensen, K. and Friis, M.L. (2001)
Genetic and environmental factors in epilepsy: a population-based study
of 11900 Danish twin pairs. Epilepsy Res., 44, 167–178.
2. Crino, P.B. (2007) Gene expression, genetics, and genomics in epilepsy:
some answers, more questions. Epilepsia, 48(Suppl. 2), 42–50.
3. Kwan, P. and Brodie, M.J. (2000) Early identification of refractory
epilepsy. N. Engl. J. Med., 342, 314–319.
4. Cavalleri, G.L., Weale, M.E., Shianna, K.V., Singh, R., Lynch, J.M.,
Grinton, B., Szoeke, C., Murphy, K., Kinirons, P., O’Rourke, D. et al.
(2007) Multicentre search for genetic susceptibility loci in sporadic
epilepsy syndrome and seizure types: a case-control study. Lancet
Neurol., 6, 970–980.
5. Kasperaviciute, D., Catarino, C.B., Heinzen, E.L., Depondt, C., Cavalleri,
G.L., Caboclo, L.O., Tate, S.K., Jamnadas-Khoda, J., Chinthapalli, K.,
Clayton, L.M. et al. (2010) Common genetic variation and susceptibility
to partial epilepsies: a genome-wide association study. Brain, 133,
2136–2147.
6. Helbig, I., Scheffer, I.E., Mulley, J.C. and Berkovic, S.F. (2008)
Navigating the channels and beyond: unravelling the genetics of the
epilepsies. Lancet Neurol., 7, 231–245.
7. Berkovic, S.F., Howell, R.A., Hay, D.A. and Hopper, J.L. (1998)
Epilepsies in twins: genetics of the major epilepsy syndromes. Ann.
Neurol., 43, 435–445.
8. Diaz-Arrastia, R., Gong, Y., Fair, S., Scott, K.D., Garcia, M.C., Carlile,
M.C., Agostini, M.A. and Van Ness, P.C. (2003) Increased risk of late
posttraumatic seizures associated with inheritance of APOE epsilon4
allele. Arch. Neurol., 60, 818–822.
9. Yamamoto, M., Yoshimura, K., Kitada, M., Nakahara, J., Seiwa, C., Ueki,
T., Shimoda, Y., Ishige, A., Watanabe, K. and Asou, H. (2009) A new
monoclonal antibody, A3B10, specific for astrocyte-lineage cells
recognizes calmodulin-regulated spectrin-associated protein 1 (Camsap1).
J. Neurosci. Res., 87, 503–513.
10. Baines, A.J., Bignone, P.A., King, M.D., Maggs, A.M., Bennett, P.M.,
Pinder, J.C. and Phillips, G.W. (2009) The CKK domain (DUF1781)
binds microtubules and defines the CAMSAP/ssp4 family of animal
proteins. Mol. Biol. Evol., 26, 2005–2014.
11. Mulle, C., Sailer, A., Perez-Otano, I., Dickinson-Anson, H., Castillo, P.E.,
Bureau, I., Maron, C., Gage, F.H., Mann, J.R., Bettler, B. et al. (1998)
Altered synaptic physiology and reduced susceptibility to kainate-induced
seizures in GluR6-deficient mice. Nature, 392, 601–605.
12. Kortenbruck, G., Berger, E., Speckmann, E.J. and Musshoff, U. (2001)
RNA editing at the Q/R site for the glutamate receptor subunits GLUR2,
GLUR5, and GLUR6 in hippocampus and temporal cortex from epileptic
patients. Neurobiol. Dis., 8, 459–468.
13. Telfeian, A.E., Federoff, H.J., Leone, P., During, M.J. and Williamson, A.
(2000) Overexpression of GluR6 in rat hippocampus produces seizures
and spontaneous nonsynaptic bursting in vitro. Neurobiol. Dis., 7,
362–374.
14. Birnbaum, S.G., Varga, A.W., Yuan, L.L., Anderson, A.E., Sweatt, J.D.
and Schrader, L.A. (2004) Structure and function of Kv4-family transient
potassium channels. Physiol. Rev., 84, 803–833.
15. Barnwell, L.F., Lugo, J.N., Lee, W.L., Willis, S.E., Gertz, S.J., Hrachovy,
R.A. and Anderson, A.E. (2009) Kv4.2 knockout mice demonstrate
increased susceptibility to convulsant stimulation. Epilepsia, 50,
1741–1751.
16. Aronica, E., Boer, K., Doorn, K.J., Zurolo, E., Spliet, W.G., van Rijen,
P.C., Baayen, J.C., Gorter, J.A. and Jeromin, A. (2009) Expression and
localization of voltage dependent potassium channel Kv4.2 in epilepsy
associated focal lesions. Neurobiol. Dis., 36, 81–95.
17. Backx, L., Ceulemans, B., Vermeesch, J.R., Devriendt, K. and Van Esch,
H. (2009) Early myoclonic encephalopathy caused by a disruption of the
neuregulin-1 receptor ErbB4. Eur. J. Hum. Genet., 17, 378–382.
18. Fazzari, P., Paternain, A.V., Valiente, M., Pla, R., Lujan, R., Lloyd, K.,
Lerma, J., Marin, O. and Rico, B. (2010) Control of cortical GABA
circuitry development by Nrg1 and ErbB4 signalling. Nature, 464,
1376–1380.
19. Parrott, A.M. and Mathews, M.B. (2007) Novel rapidly evolving hominid
RNAs bind nuclear factor 90 and display tissue-restricted distribution.
Nucleic Acids Res., 35, 6249–6258.
20. Lauren, J., Airaksinen, M.S., Saarma, M. and Timmusk, T. (2003) A novel
gene family encoding leucine-rich repeat transmembrane proteins
differentially expressed in the nervous system. Genomics, 81, 411–421.
21. Fuerst, P.G., Bruce, F., Tian, M., Wei, W., Elstrott, J., Feller, M.B.,
Erskine, L., Singer, J.H. and Burgess, R.W. (2009) DSCAM and
DSCAML1 function in self-avoidance in multiple cell types in the
developing mouse retina. Neuron, 64, 484–497.
22. Amano, K., Fujii, M., Arata, S., Tojima, T., Ogawa, M., Morita, N.,
Shimohata, A., Furuichi, T., Itohara, S., Kamiguchi, H. et al. (2009)
DSCAM deficiency causes loss of pre-inspiratory neuron synchroneity
and perinatal death. J. Neurosci., 29, 2984–2996.
23. Ma, D.K., Chiang, C.H., Ponnusamy, K., Ming, G.L. and Song, H. (2008)
G9a and Jhdm2a regulate embryonic stem cell fusion-induced
reprogramming of adult neural stem cells. Stem Cells, 26, 2131–2141.
24. Ostrowski, L.E., Andrews, K., Potdar, P., Matsuura, H., Jetten, A. and
Nettesheim, P. (1999) Cloning and characterization of KPL2, a novel gene
induced during ciliogenesis of tracheal epithelial cells. Am. J. Respir. Cell.
Mol. Biol., 20, 675–683.
25. Dichter, M.A. (2009) Emerging concepts in the pathogenesis of epilepsy
and epileptogenesis. Arch. Neurol., 66, 443–447.
26. Berg, A.T., Berkovic, S.F., Brodie, M.J., Buchhalter, J., Cross, J.H., van
Emde Boas, W., Engel, J., French, J., Glauser, T.A., Mathern, G.W. et al.
(2010) Revised terminology and concepts for organization of seizures and
epilepsies: report of the ILAE Commission on Classification and
Terminology, 2005–2009. Epilepsia, 51, 676–685.
27. Price, A.L., Patterson, N.J., Plenge, R.M., Weinblatt, M.E., Shadick, N.A.
and Reich, D. (2006) Principal components analysis corrects for
stratification in genome-wide association studies. Nat. Genet., 38,
904–909.
28. Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M.A.,
Bender, D., Maller, J., Sklar, P., de Bakker, P.I., Daly, M.J. et al. (2007)
PLINK: a tool set for whole-genome association and population-based
linkage analyses. Am. J. Hum. Genet., 81, 559–575.
29. Higgins, J.P. and Thompson, S.G. (2002) Quantifying heterogeneity in a
meta-analysis. Stat. Med., 21, 1539–1558.
30. Howie, B.N., Donnelly, P. and Marchini, J. (2009) A flexible and accurate
genotype imputation method for the next generation of genome-wide
association studies. PLoS Genet., 5, e1000529.
6 Human Molecular Genetics, 2011
at Head Serials and Electronic Resources Dept on January 6, 2012
http://hmg.oxfordjournals.org/
Downloaded from