Gender Differences in Associations of Glutamate
Decarboxylase 1 Gene (GAD1) Variants with Panic
Heike Weber1*, Claus Ju ¨rgen Scholz2, Katharina Domschke1,3, Christian Baumann1, Benedikt Klauke3,
Christian P. Jacob1, Wolfgang Maier4, Ju ¨rgen Fritze5, Borwin Bandelow6, Peter Michael Zwanzger3,
Thomas Lang7,8, Lydia Fehm9, Andreas Stro ¨hle10, Alfons Hamm11, Alexander L. Gerlach12,
Georg W. Alpers13,14, Tilo Kircher15, Hans-Ulrich Wittchen16, Volker Arolt1, Paul Pauli13, Ju ¨rgen Deckert1,
1Department of Psychiatry, Psychosomatics and Psychotherapy, University of Wu ¨rzburg, Wu ¨rzburg, Germany, 2Interdisciplinary Center for Clinical Research (IZKF),
University of Wu ¨rzburg, Wu ¨rzburg, Germany, 3Department of Psychiatry and Psychotherapy, University of Mu ¨nster, Mu ¨nster, Germany, 4Department of Psychiatry,
University of Bonn, Bonn, Germany, 5Department of Psychiatry, University of Frankfurt, Frankfurt, Germany, 6Department of Psychiatry, University of Go ¨ttingen,
Go ¨ttingen, Germany, 7Christoph-Dornier-Foundation for Clinical Psychology, Bremen, Germany, 8Zentrum fu ¨r Klinische Psychologie und Rehabilitation, University of
Bremen, Bremen, Germany, 9Institute of Psychology, University of Berlin, Berlin, Germany, 10Department of Psychiatry, Charite ´, Berlin, Germany, 11Institute of
Psychology, University of Greifswald, Greifswald, Germany, 12Institute of Clinical Psychology and Psychotherapy, University of Ko ¨ln, Ko ¨ln, Germany, 13Department of
Psychology, University of Wu ¨rzburg, Wu ¨rzburg, Germany, 14Clinical and Biological Psychology, School of Social Sciences, University of Mannheim, Mannheim,
Germany, 15Department of Psychiatry, University of Marburg, Marburg, Germany, 16Institute of Clinical Psychology and Psychotherapy, Technical University of Dresden,
Background: Panic disorder is common (5% prevalence) and females are twice as likely to be affected as males. The
heritable component of panic disorder is estimated at 48%. Glutamic acid dehydrogenase GAD1, the key enzyme for the
synthesis of the inhibitory and anxiolytic neurotransmitter GABA, is supposed to influence various mental disorders,
including mood and anxiety disorders. In a recent association study in depression, which is highly comorbid with panic
disorder, GAD1 risk allele associations were restricted to females.
Methodology/Principal Findings: Nineteen single nucleotide polymorphisms (SNPs) tagging the common variation in
GAD1 were genotyped in two independent gender and age matched case-control samples (discovery sample n=478;
replication sample n=584). Thirteen SNPs passed quality control and were examined for gender-specific enrichment of risk
alleles associated with panic disorder by using logistic regression including a genotype6gender interaction term. The latter
was found to be nominally significant for four SNPs (rs1978340, rs3762555, rs3749034, rs2241165) in the discovery sample;
of note, the respective minor/risk alleles were associated with panic disorder only in females. These findings were not
confirmed in the replication sample; however, the genotype6gender interaction of rs3749034 remained significant in the
combined sample. Furthermore, this polymorphism showed a nominally significant association with the Agoraphobic
Cognitions Questionnaire sum score.
Conclusions/Significance: The present study represents the first systematic evaluation of gender-specific enrichment of risk
alleles of the common SNP variation in the panic disorder candidate gene GAD1. Our tentative results provide a possible
explanation for the higher susceptibility of females to panic disorder.
Citation: Weber H, Scholz CJ, Domschke K, Baumann C, Klauke B, et al. (2012) Gender Differences in Associations of Glutamate Decarboxylase 1 Gene (GAD1)
Variants with Panic Disorder. PLoS ONE 7(5): e37651. doi:10.1371/journal.pone.0037651
Editor: Kenji Hashimoto, Chiba University Center for Forensic Mental Health, Japan
Received January 24, 2012; Accepted April 23, 2012; Published May 25, 2012
Copyright: ? 2012 Weber et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This study was supported by the Deutsche Forschungsgemeinschaft(DFG/German Reasearch Funding Organization; Grant SFB TRR 58 Z02 to AR, PP
and JD, C01 to PZ and C02 to KD; RE1632/1-5 to AR, KFO 125 to AR, PP and CPJ; RTG 1252, to AR, PP and JD; for details see website http://www.dfg.de) and the
Bundesministerium fu ¨r Bildung und Forschung (Federal Ministry of Education and Research/BMBF; Panic-Net, to ALG, TK, AS, HUW, VA, AH and JD; for details see
web pages http://www.bmbf.de/ and http://www.paniknetz.de/netzwerk.html). The funders had no role in study design, data collection and analysis, decision to
publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: Weber_H2@klinik.uni-wuerzburg.de
The lifetime prevalence of anxiety disorders has been estimated
to be as high as 30% with a spectrum that ranges from 2% for
agoraphobia up to 12% for specific phobia; with the prevalence of
panic disorder being estimated at 5% . Of note, the prevalence
of panic disorder is twice as high in females as in males . Panic
disorder displays a high comorbidity with other anxiety (e.g.
PLoS ONE | www.plosone.org1May 2012 | Volume 7 | Issue 5 | e37651
agoraphobia) and mood disorders (e.g. unipolar depression), which
phenotypically share the personality trait neuroticism [3,4].
Current drug treatment involves the use of benzodiazepines,
whose anxiolytic effect is caused by increasing the responsiveness
of the c-aminobutyric acid (GABA) receptor . Correspondingly,
the results of several magnetic resonance spectroscopy studies
point to an impaired GABA system in panic disorder patients
[6,7]. The key enzyme for the synthesis of the inhibitory
neurotransmitter GABA is glutamic acid dehydrogenase (GAD)
which uses glutamate and pyridoxal phosphate as substrates for
the reaction that takes place in presynaptic neurons (reviewed in
); it exists in two isoforms – 67 kDa and 65 kDa – encoded by
different genes (GAD1, located at 2q31.1 and GAD2 (10p11.23),
respectively). Intriguingly, brain GAD expression was shown to be
influenced by sex hormones in female rats and rhesus macaques
[9,10]. In the latter species, estrogen and progesterone decrease
GAD expression in the amygdala and the hippocampus (which
both are involved in regulating fear), which provides a link
between hormone levels and anxiety as well as mood attacks
during menstruation in humans .
A recent twin study estimated the heritability of panic disorder
at 48%  and an independently conducted genome-wide linkage
scan yielded susceptibility loci at chromosomal regions 2q
(including GAD1) and 15q . With the assumption that both
GAD genes may serve as plausible candidates to influence
neuroticism, the examination of allelic variation in these genes
revealed associations of GAD1 single nucleotide polymorphisms
(SNPs) rs2241165, rs2058725 and rs3791850 in a mixed anxiety
(including panic) and mood disorder sample . Studies in
bipolar disorder further corroborate the association of GAD1 with
mood disorders (rs1978340, rs872123 and rs2241165) [13,14], but
also in schizophrenia, GAD1 alleles were shown to convey genetic
risk (rs10432420, rs3749035, rs16823181, rs3791878, rs3791858,
rs3749034, rs2270335, rs2241165, rs379850) [15,16,17,18]. Re-
cently, the GAD1 SNPs examined in  were (along with SNPs
mapping to other genes) explored in an independent sample of
patients suffering from major depression, with a particular focus
on sleep disturbance subtypes and gender . Those GAD1 SNPs
that were not (rs12185692) or only marginally (rs769407)
associated with neuroticism  revealed female-specific associa-
tions with depression . Since depression as well as panic
disorder are both more prevalent among females, this striking
gender difference prompted us to investigate whether gender-
specific associations of GAD1 variants are also detectable in panic
disorder. In the present study we used two independent gender-
matched case/control samples with a total size of ncase=531 to test
this hypothesis. Nominally significant gender differences and
associations with disease were found in the discovery sample,
which were however not supported by the replication sample; in
the analysis of the combined sample, only one SNP displayed a
nominally significant gender-specific risk allele enrichment, the
reasons for which are discussed in depth below. Studies in larger
samples and subsequent meta-analysis are thus needed to
unequivocally prove a possible female-specific contribution of
the GAD1 gene in panic disorder.
In order to capture the common allelic variation in GAD1 and
approximately 10 kb of the gene’s up- and downstream flanking
regions, we chose a representative set of 19 SNPs (based on
HapMap CEU data) for genotyping. Thereof, six SNPs did not
pass our stringent quality criteria in the discovery sample; hence,
the remaining set of 13 SNPs (see Table S1) was still found to be
representative for .80% of common HapMap SNPs in the
defined region. In a first analysis step, we tested our a priori
hypothesis by modeling panic disorder disease risk with the
number of risk alleles conditioning on gender, implementing a
multiple regression including genotype6gender interactions in
statistical terms. The odds ratio (OR) derived from this interaction
term represents the ratio of effect sizes (i.e. ORs) from genetic
associations in males versus those in females, thus quantifying
differential risk allele enrichment between genders. In the
discovery sample, four SNPs displayed nominally significant
genotype6gender interactions; particularly the respective minor
alleles (assumed to convey the genetic risk) of rs3762555,
rs3749034 and rs2241165 had effects that were almost twice as
large in males as in females (ORs close to 2) and vice versa for
rs1978340 (OR close to 0.5; see Table 1). These four candidate
SNPs were subjected to a post hoc analysis in which we tested the
genotype associations with panic disorder separately in each
gender; this aimed at investigating whether these nominally
significant gender interactions go along with SNP associations in
both genders, however with opposing effects. Indeed, the
estimated effect sizes for the four SNPs were opposite between
genders (i.e. risk/OR.1 versus protective/OR,1), however,
associations reached nominal significance only in females (see
Table 2). Notably, the associations of rs3762555, rs3749034 and
rs2241165 are unlikely to be independent because these SNPs are
located on the same haplotype block (see Figure S1).
In order to replicate these initial findings, the 13 SNPs
examined in the discovery sample were genotyped in an
independent replication sample; thereof, genotyping results of
rs4439928 did not meet the quality criteria and accordingly this
SNP was excluded from replication analysis. The overall linkage
disequilibrium structure of the replication sample was found to be
similar to the discovery sample (see Figure S2). However, in
contrast to the results from the discovery sample, none of the
gender-specific risk allele enrichments reached even marginal
(p,0.1) significance (see Table 1). Furthermore, none of the four
Table 1. Genotype6gender interaction analysis of common
GAD1 alleles associated with panic disorder.
rs ID minor/majorORpORpORp
rs1978340 A/G0.519 0.032 1.2920.4060.8320.390
rs3762555C/G 1.872 0.041 1.4460.2791.4960.068
rs3749034A/G1.992 0.023 1.4450.2912.009 0.045
rs2241165C/T1.958 0.030 1.3250.4111.4830.076
rs701492T/C 0.7450.3380.909 0.7600.8280.380
rs16858996 G/A0.6240.2941.565 0.4120.9650.914
*rs4439928 did not meet the quality criteria in the replication sample.
Nominally significant results are shown in bold. Abbreviations: OR, odds ratio; p,
p-value; SNP, single nucleotide polymorphism.
Associations of GAD1 with Panic Disorder
PLoS ONE | www.plosone.org2May 2012 | Volume 7 | Issue 5 | e37651
candidate SNPs showed the trend to be associated with panic
disorder in gender-specific subsets of the replication sample (see
Table 2). Despite this apparently failed replication, it should be
noted that the number of male probands (n=152) as well as the
male proportion of the replication sample (26%) are smaller than
those of the discovery sample (n=192, 40%). In order to exclude
false positive as well as false negative associations as a result from
insufficient sampling, we decided to further analyze the combined
sample. In this setting, three of four significant gender differences
found in the discovery sample gained at least marginal significance
(see Table 1). However, the four candidate SNPs were not
associated with panic disorder in gender subsets (see Table 2).
Since the replication sample offers the advantage of known
dimensional phenotypes, we furthermore aimed to quantify the
effect of our candidate SNPs on ASI and ACQ sum scores. The
range of these scores differs between healthy individuals and panic
patients per definition, resulting in unequal variances between
groups. We therefore determined gender-wise genotype associa-
tions with either sum score separately for the panic and the control
groups. In all but one case, the estimated effects conveyed by each
risk allele were smaller in males compared to females, however,
only two estimated effects were significantly different from zero
(see Table 3). Although largely insignificant, these results lend
qualitative support to our hypothesis that GAD1 alleles may have a
different impact on panic disorder susceptibility in both genders.
In addition to genetic association testing we tried to predict
functional consequences of our candidate SNPs using a bioinfor-
matic approach. We found that SNPs close to GAD1’s transcrip-
tion start site are linked to putative changes of promoter function.
Particularly, the minor allele of rs1978340 (upstream) deletes a
putative transcription factor binding site (TFBS) for the E2F4-
TFDP2 dimer, whereas the minor alleles of rs3762555 (upstream)
and rs3749034 (59 untranslated region) create predicted TFBSs for
SRF and ZEB1/ZFHEP/AREB6, respectively. For the intronic
SNP rs2241165 no allele-specific functional consequences (affect-
ing e.g. splicing) were evident, nor is linkage disequilibrium data
available for nearby SNPs with putative structural impact
mediated by amino acid exchanges (rs77655188/GAD1T27K;
rs17857148/GAD1H291R; rs17857149/GAD1K318E). The restriction
of this analysis to our candidate SNPs imposes the limit that true
causal variants may be missed, since each of our SNPs examined
in this study is representative for the variation of several
the mode of action of gender-wise differentially enriched GAD1
risk alleles affects promoter function and thus transcriptional
In this paper, we present the first systematic evaluation of
gender-dependent associations of common SNP variation in GAD1
with panic disorder. However, the initial findings from our clinical
discovery sample were not supported by our replication sample
originating from a cognitive behavioral psychotherapy (CBT) trial.
Of four genotype associations, no replicated association was found
at the categorical level and in the combined sample only the
Table 2. Gender-wise association of GAD1 polymorphisms displaying gender-specific risk allele enrichment with panic disorder.
rs ID minor/major ORp ORp ORp
rs1978340A/G female 1.5880.012 1.0040.978 1.2210.091
male 0.8250.429 1.2980.326 1.0160.929
rs3762555 C/Gfemale0.648 0.030 1.0820.614 0.8880.330
male 1.2130.407 1.5650.139 1.328 0.122
rs3749034 A/Gfemale0.623 0.0194.066 0.7530.862 0.222
male1.242 0.336 1.5180.1811.335 0.110
rs2241165C/T female 0.6150.016 1.1060.5200.8810.299
male 1.205 0.4271.465 0.2101.306 0.15
Nominally significant results are shown in bold. Abbreviations: OR, odds ratio; p, p-value; SNP, single nucleotide polymorphism.
Table 3. Associations of GAD1 candidate SNPs with
dimensional anxiety traits.
rs IDminor/majorincrease pincrease p
ACQ rs1978340 A/Gfemale0.0430.5120.008 0.782
rs3749034 A/G female
2 20.263 0.027 20.007 0.868
ASIrs1978340 A/Gfemale1.8910.130 0.0940.902
2 22.641 0.015
Nominally significant results are shown in bold. Abbreviations: ACQ,
Agoraphobic Cognitions Questionnaire; ASI, Anxiety Sensitivity Index; OR, odds
ratio; p, p-value; SNP, single nucleotide polymorphism.
Associations of GAD1 with Panic Disorder
PLoS ONE | www.plosone.org3May 2012 | Volume 7 | Issue 5 | e37651
interaction of rs3749034 with gender became significant. Inter-
estingly though, rs3749034 was also among the SNPs associated
with the dimensional sum score of ACQ in the replication sample,
but at first sight this dimensional association seems to contradict
the categorical results: although the latter revealed significant
gender-differences, rs3749034 had a protective effect on all
females in the discovery sample and on male panic patients in
the replication sample as well. However, since the ACQ quantifies
agoraphobia , the analyses of our dimensional and categorical
phenotypes need not necessarily reveal concordant results. This
point is of particular interest because the discovery and replication
samples differ in their comorbidity with agoraphobia (see below
and Table S2). Finally, a concordantly significant reduction of
ACQ in male replication controls was also not detectable, arguing
for the possibility that the association of rs3749034 with ACQ in
male panic patients may be a false positive.
A major limitation of this study is imposed by the restricted
sample size and the high female to male ratio. Due to the high LD
within the examined region, we used nominal p-values for the
assessment of significance to avoid the (in this case) overly strict
Bonferroni correction, which revealed associations in the discovery
sample that would otherwise have been negative. This measure
decreases type-II error, while increasing type-I error, which has to
be taken into account when interpreting the data. Due to the lower
prevalence of panic disorder in males, especially the male
subsamples were rather limited in size, which may hinder the
realistic estimation of genetic effect sizes because of insufficient
sampling and therefore confound possible gender differences in
risk allele enrichment. This point is more relevant to the
replication sample, which features an even smaller male sub-
sample as compared to the discovery sample, and which did not
support the findings from the discovery sample. As the combined
sample did not reflect our associations from the discovery sample
either, it is tempting to speculate that lack of replication may also
be explained by differences in the composition of our samples (see
Table S2): Firstly, mean age of discovery cases (37.59611.13) and
controls (36.18611.78) were more similar in the discovery than in
the replication sample(cases,
28.867.38). Due to the younger age of replication controls, some
of the apparently healthy controls at the time of blood donation
may not have reached the age of onset for panic disorder, so that
the fraction of persons predisposed to panic disorder may be
higher in replication than in discovery controls . On the other
hand, replication controls were screened against mental axis 1
disorders, whereas discovery controls were anonymous blood
donors. Nevertheless the likelihood of severe psychiatric disorders
present in the control sample is assumed to be low, because blood
donors were informed about the intended enrollment in this study.
Therefore, given the panic disorder prevalence of 5% , in a
worst-case scenario the replication controls (n=292) may contain
up to 15 panic disorder patients, which should result in only a
minor distortion of the OR. Secondly, the patients of the discovery
sample were a mixed in- and outpatient sample from psychiatric
departments recruited with the primary diagnosis of panic
disorder, while the patients of the replication sample were all
outpatients from psychological outpatient services volunteering for
a CBT trial of panic with agoraphobia and thus recruited with the
primary diagnosis panic disorder with agoraphobia. Of the 239
panic patients of our discovery sample, only 68.6% (n=164) thus
were diagnosed with comorbid agoraphobia, whereas this was the
case for all 292 panic patients of the replication sample. Despite
the high degree of comorbidity of panic disorder and agoraphobia,
the genetic architecture may differ between the two samples. This
notion is supported by results from association tests with swapped
samples: discovery cases versus replication cases again revealed
significant differences in some of the associated markers of the
discovery sample; discovery controls and replication controls
however were not different. On the other hand, discovery cases
versus replication controls partly confirmed the discovery findings,
but not replication cases versus discovery controls (see Table S3).
Thus, the observed gender differences may be caused by ‘‘pure’’
panic patients in the discovery sample (n=75). Further explor-
atory analyses however revealed a trend for gender-specific
associations in sub-samples of panic disorder with and without
comorbid agoraphobia (see Table S4).
In conjunction with the present genetic data, convergent lines of
evidence suggest that GAD1 and its gene product are implicated in
the pathophysiology of panic disorder. GABA levels in various
brain regions are reduced in panic patients [6,7], possibly due to
impaired GAD function. Further studies in patients with major
depression, a mechanistically related disorder, found reduced
GABA levels to be accompanied by increased glutamate concen-
trations , strengthening the link between anxiety and mood
disorders and GAD. In patients with stiff person syndrome (SPS)
GAD function is decreased due to the development of auto-
antibodies against this enzyme; these are not only responsible for
the primary phenotype – increased muscle stiffness and intermit-
tent muscle spasms – but also raise anxiety in SPS patients [22,23].
This connection is particularly supported by the fact that passive
transfer of GAD antibodies to healthy rats resulted in an anxious
phenotype in these animals was well . These observations are
complemented with genetic evidence, which shows that GAD1
SNPs are associated with neuroticism (rs2241165, rs2058725 and
rs3791850) and anxiety (rs769407, rs3791851 and rs769395)
[12,25]. Furthermore, the GAD1 polymorphism rs1978340 was
found to influence brain GABA levels: healthy individuals that are
homozygous for the minor allele display higher concentrations
than major allele carriers . Intriguingly, this seems to
contradict our result that rs1978340 is associated with increased
risk for panic disorder in females (see Table 2). However, this
particular SNP’s effect on the risk to be affected by schizophrenia,
a disorder that also features reduced brain GABA levels ,
seems to depend on the COMT promoter polymorphism
rs2075507 , so that epistasis seems to play an important role.
This however remains elusive with respect to GABA concentra-
Further genetic studies on psychiatric disorders are in line with
our tentative finding of gender-specific enrichment of GAD1 risk
alleles in panic disorder. In the latter, a linkage analysis of an US
American sample revealed that both chromosome 2 microsatellite
marker loci, which flank the GAD1 gene at positions 169 and
178 cM, display moderate differences in male and female
associations . In unipolar depression, which is comorbid with
panic disorder, a Finnish association study found alleles of the
GAD1 polymorphisms rs12185692 (upstream) and rs769407 (in the
sixth intron) to go along with disease only in females .
Furthermore, several studies on schizophrenia also revealed
gender differences in associations of GAD1 SNPs [15,16,17]. An
overview of gender-specific SNP associations is given in Table S5.
In line with this, other association studies examining GAD1 in
mixed gender samples of depression , schizophrenia  and
autism [30,31] revealed no associations. This also holds true for
the examination of our samples without subgrouping by gender
(see Table S6). On the other hand, the studies on neuroticism and
anxiety demonstrated GAD1 SNP associations even without
considering gender effects [12,25]. Compared to our discovery
sample, the latter studies employed a larger sample size, which
increases the power to detect smaller genetic effects.
Associations of GAD1 with Panic Disorder
PLoS ONE | www.plosone.org4 May 2012 | Volume 7 | Issue 5 | e37651
Since the heritability for panic disorder is somewhat lower than
that for other mental disorders [32,33], non-genetic (i.e. environ-
mental) influences contribute to a large part to the etiology of
panic disorder. One important class of environmental influences in
terms of panic disorder are life events . Recently, in an
experimental paradigm it was shown that low maternal care –
which might serve as an animal experimental model for early life
stress in humans – leads to increased CpG methylation as well as
to decreased H3K9 acetylation of the Gad1 promoter in the rat
hippocampus, thus silencing gene expression . It is therefore
conceivable that life events and underlying molecular mechanisms
like e.g. DNA methylation and histone acetylation may interfere
with genetic effects. Such a gene6environment correlation of
GAD1 SNPs and childhood adversities has very recently been
examined in the context of anxiety disorders, however failed to
detect significant effects . In the light of hormone-induced
gender-dependent GAD1 expression  however, a gene6envir-
onment6gender correlation seems plausible which has not been
tested for in this study.
In summary, our study provides supporting evidence for gender
differences in the role of GAD1 variation for the pathogenesis of
panic disorder, but is far from conclusive. Future studies in larger
gender-balanced samples well characterized for panic disorder and
agoraphobia as well as comorbid psychopathology (including
dimensional phenotypes) and life events will be necessary to
evaluate the relevance of our tentative findings.
Materials and Methods
All SNPs examined in this study were genotyped in panic
disorder (with or without agoraphobia) patients which were
matched by gender to an equal number of healthy controls.
Details on case/control pairs of the discovery and the replication
sample are specified below. All cases as well as controls were
unrelated and of self-reported Western European descent. Patients
with comorbid schizoaffective or other psychotic disorders,
substance abuse disorders, mental retardation, neurological or
neurodegenerative disorders were excluded. Only patients and
volunteers who gave written informed consent were enrolled in the
study. The present study complied with the Declaration of
Helsinki and was specifically approved by the Ethics Committee
of the University Hospital, University of Wu ¨rzburg. A demo-
graphic overview of the discovery and the replication samples can
be found in Table S2.
The discovery sample consisted of 239 panic disorder patients,
collected in Bonn, Wu ¨rzburg, Mu ¨nster and Go ¨ttingen (fe-
male=143, male=96; mean age 37.59611.13), from out- and
inpatients of the respective centers treated there as part of the
regular clinical care. The diagnosis of panic disorder and absence
or presence of comorbid agoraphobia (68.6%; n=164) was
ascertained by experienced clinicians on the basis of medical
records and structured or standardized clinical interviews (Sched-
ule for Affective Disorders and Schizophrenia (lifetime version),
SADS-LA; Structured Clinical Interview for DSM IV, SCID; and
Composite Inernational Diagnostic Interview, CIDI) according to
the criteria of DSM- (Diagnostic and Statistical Manual of Mental
Disorders) III-R or DSM-IV, respectively [36,37,38]. The gender-
matched control sample comprised 239 anonymous blood donors
36.18611.78) that were not screened for psychiatric disorders.
However, all apparently healthy individuals were aware of their
intended enrollment in the control sample of the study and
furthermore, as a requirement for blood donation, were free of
medication. Therefore the likelihood of severe psychiatric disor-
ders present in the control sample was assumed to be low.
The replication sample’s patients (n=292; female=216,
male=76; mean age 36.04610.77) were enrolled from the BMBF
‘‘Panic-Net’’ multicenter psychotherapy treatment study; 15 of these
patients did not enter the study yet fulfilled the respective inclusion
criteria. Patients were enrolled from specialized outpatient
treatment units within the framework of the Panic-Net study as
described in detail . The diagnosis of panic disorder with
agoraphobia was established by a standardized clinical interview
(CIDI) according to DSM-IV criteria . Controls used for
genotypic associations with panic disorder (n=292; female=216,
male=76; mean age 28.867.38) were from Mu ¨nster and
Wu ¨rzburg, matched by gender to cases and drawn from a larger
number of screened healthy controls (n=1564; female=809,
male=755; mean age 24.9365.11 ascertained within the frame-
work of the Collaborative Research Center TRR SFB 58).
Absence of mental axis 1 disorders was established by experienced
psychologists on the basis of a structured clinical interview (Mini
International Neuropsychiatric Interview, MINI) according to the
criteria of DSM-IV . For both patient and control groups,
panic fear and anxiety sensitivity were evaluated by German
versions of the Anxiety Sensitivity Index (ASI, ) and
Agoraphobic Cognitions Questionnaire (ACQ, ).
In order to capture allelic variation in the GAD1 gene, 19 tag
SNPs were derived from HapMap CEU data  using the Tagger
function as implemented in Haploview V4.2  with default
settings. SNP genotyping was performed using Sequenom’s
MassArrayH system according to the instructions supplied by the
manufacturer. All PCR reactions were done with the iPlexH
chemistry following the MassArrayH iPlexH standard operation
procedure. Primer sequences can be found in Table S7.
Prior to statistical analysis, SNPs had to fulfill several quality
control criteria. The minimal call rate threshold was set to 90%,
SNPs with a call rate below this threshold (rs769390, rs4668331
and rs4439928, the latter only in the replication sample) where not
further analyzed. Deviations from Hardy-Weinberg equilibrium
(HWE) were considered to be indicative for the presence of
genotyping errors; SNPs rs769393 and rs769395 yielded a HWE
p-value below the threshold of 0.01 and were thus excluded from
further analysis. Also, rare variants with a minor allele frequency
(MAF) below 5% (rs769406, rs12472230) were not further
examined. The resulting set of 13 SNPs tagging the common
allelic variation in the GAD1 gene was analyzed with logistic
regression, assuming the minor alleles to convey the genetic risk in
an additive manner. SNP associations with dimensional anxiety
traits were analyzed with univariate linear regressions modeling
the ACQ and the ASI sum scores in each gender separately. Each
risk allele’s effect is presented as OR for categorical outcomes or as
linear increase for dimensional phenotypes. The respective p-
values represent the probabilities for the estimated effects being
indistinguishable from zero. P-values ,0.05 are termed nominally
significant. Although conducting repeated statistical tests, p-values
were not corrected for multiple testing, which was – despite large
sample size - due to the limited power of the study. Given a SNP
with MAF=5% conveying a moderate relative risk of 1.5 to
develop panic disorder in an additive model, the power to find a
nominal association (p,0.05) is 65% for the mixed gender
combined sample (n=1062), 48% for its female (n=718) and 26%
for its male (n=344) subset. In the same analysis setting, but split
Associations of GAD1 with Panic Disorder
PLoS ONE | www.plosone.org5May 2012 | Volume 7 | Issue 5 | e37651
up in discovery and replication sample, the mixed gender
discovery sample (n=478) and its female (n=286) and male
(n=192) subsets provide 35%, 24% and 18% power, respectively;
the mixed gender replication sample (n=584) and its female
(n=432) and male (n=152) subsets provide 42%, 32% and 15%
power, respectively. The analyses were performed with Haploview
V4.2 , PGA , PLINK V1.07  and R V2.11 .
Assessment of SNP function
Analyses of SNPs and their sequence contexts were performed
with tools that are contained in the GenEpi toolbox .
Differential TFBS predictions were made using the web-based
tool MatInspector . Summary tables produced by F-SNP
[48,49] were used to indicate a SNP’s possible influence on
GAD1 single nucleotide polymorphisms examined in
the discovery sample. LD analysis was performed with
Haploview v4.2 using default settings, i.e. D9 was used as measure
for LD and haplotype blocks were defined with the method
Linkage disequilibrium (LD) structure of
GAD1 single nucleotide polymorphisms examined in
the replication sample. LD analysis was performed with
Haploview v4.2 using default settings, i.e. D9 was used as measure
for LD and haplotype blocks were defined with the method
Linkage disequilibrium (LD) structure of
amined in the study. rs4439928 did not meet the quality
criteria in the replication sample and was therefore not included in
the replication and combined sample analysis.
Single nucleotide polymorphisms (SNPs) ex-
the replication sample. Abbreviations: AG, agoraphobia;
CIDI, Composite International Diagnostic Interview; DSM,
Diagnostic and Statistical Manual of Mental Disorders; MINI,
Mini International Neuropsychiatric Interview; SADS-LA, Sched-
ule for Affective Disorders and Schizophrenia (lifetime version);
SCID, Structured Clinical Interview for DSM IV.
Demographic overview of the discovery and
morphisms in swapped samples. Analysis of gender
Gender-specific associations of GAD1 poly-
subsamples was performed only for polymorphisms that displayed
significant gender differences in the discovery case-control sample
(see Table 1); rs4439928 did not meet the quality criteria in the
replication sample and was therefore not included in the swapped
sample analysis. Nominally significant results are shown in bold.
Abbreviations: Ca, case group; Co, control group; D, discovery
sample; OR, odds ratio; p, p-value; R, replication sample; SNP,
single nucleotide polymorphism.
morphisms in samples with and without co-morbid
agoraphobia. Analysis of gender subsamples was performed
only for polymorphisms that displayed significant gender differ-
ences in the discovery case-control sample (see Table 1);
rs4439928 did not meet the quality criteria in the replication
sample and was therefore not included in the analysis of the
combined sample. Nominally significant results are shown in bold.
Abbreviations: AG, agoraphobia; D, discovery sample; OR, odds
ratio; p, p-value; R, replication sample; SNP, single nucleotide
Gender-specific associations of GAD1 poly-
GAD1 single nucleotide polymorphisms (SNP).
Published gender-specific associations of
panic disorder in mixed-gender samples. rs4439928 did
not meet the quality criteria in the replication sample and was
therefore not included in the replication and combined mixed-
gender sample analysis. Abbreviations: OR, odds ratio; p, p-value;
SNP, single nucleotide polymorphism.
Associations of GAD1 polymorphisms with
Primer sequences used in this study.
We are grateful to all patients and healthy individuals for their
participation in the study. We thank U. Walter and M. Zimmer for their
kind help in operating the mass spectrometer. T. To ¨pner, N. Steigerwald,
C. Gagel and J. Auer are credited for excellent technical assistance.
Conceived and designed the experiments: AR JD PP. Performed the
experiments: HW. Analyzed the data: HW CJS CB BK. Contributed
reagents/materials/analysis tools: KD CPJ WM JF BB PZ TL LF AS AH
ALG GWA TK HUW VA PP JD. Wrote the paper: HW AR JD.
1. Kessler RC, Berglund P, Demler O, Jin R, Merikangas KR, et al. (2005)
Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the
National Comorbidity Survey Replication. Arch Gen Psychiatry 62: 593–602.
Bekker MH, van Mens-Verhulst J (2007) Anxiety disorders: sex differences in
prevalence, degree, and background, but gender-neutral treatment. Gend Med 4
Suppl B: S178–193.
Hettema JM, Neale MC, Myers JM, Prescott CA, Kendler KS (2006) A
population-based twin study of the relationship between neuroticism and
internalizing disorders. Am J Psychiatry 163: 857–864.
Middeldorp CM, Cath DC, Van Dyck R, Boomsma DI (2005) The co-
morbidity of anxiety and depression in the perspective of genetic epidemiology.
A review of twin and family studies. Psychol Med 35: 611–624.
Zwanzger P, Rupprecht R (2005) Selective GABAergic treatment for panic?
Investigations in experimental panic induction and panic disorder. J Psychiatry
Neurosci 30: 167–175.
6. Goddard AW, Mason GF, Almai A, Rothman DL, Behar KL, et al. (2001)
Reductions in occipital cortex GABA levels in panic disorder detected with 1h-
magnetic resonance spectroscopy. Arch Gen Psychiatry 58: 556–561.
Ham BJ, Sung Y, Kim N, Kim SJ, Kim JE, et al. (2007) Decreased GABA levels
in anterior cingulate and basal ganglia in medicated subjects with panic disorder:
a proton magnetic resonance spectroscopy (1H-MRS) study. Prog Neuropsy-
chopharmacol Biol Psychiatry 31: 403–411.
Cherlyn SY, Woon PS, Liu JJ, Ong WY, Tsai GC, et al. (2010) Genetic
association studies of glutamate, GABA and related genes in schizophrenia and
bipolar disorder: a decade of advance. Neurosci Biobehav Rev 34: 958–977.
McCarthy MM, Kaufman LC, Brooks PJ, Pfaff DW, Schwartz-Giblin S (1995)
Estrogen modulation of mRNA levels for the two forms of glutamic acid
decarboxylase (GAD) in female rat brain. J Comp Neurol 360: 685–697.
10. Noriega NC, Eghlidi DH, Garyfallou VT, Kohama SG, Kryger SG, et al. (2010)
Influence of 17beta-estradiol and progesterone on GABAergic gene expression
in the arcuate nucleus, amygdala and hippocampus of the rhesus macaque.
Brain Res 1307: 28–42.
Associations of GAD1 with Panic Disorder
PLoS ONE | www.plosone.org6 May 2012 | Volume 7 | Issue 5 | e37651
11. Fyer AJ, Hamilton SP, Durner M, Haghighi F, Heiman GA, et al. (2006) A Download full-text
third-pass genome scan in panic disorder: evidence for multiple susceptibility
loci. Biol Psychiatry 60: 388–401.
12. Hettema JM, An SS, Neale MC, Bukszar J, van den Oord EJ, et al. (2006)
Association between glutamic acid decarboxylase genes and anxiety disorders,
major depression, and neuroticism. Mol Psychiatry 11: 752–762.
13. Geller B, Tillman R, Bolhofner K, Hennessy K, Cook EH, Jr. (2008) GAD1
single nucleotide polymorphism is in linkage disequilibrium with a child bipolar I
disorder phenotype. J Child Adolesc Psychopharmacol 18: 25–29.
14. Lundorf MD, Buttenschon HN, Foldager L, Blackwood DH, Muir WJ, et al.
(2005) Mutational screening and association study of glutamate decarboxylase 1
as a candidate susceptibility gene for bipolar affective disorder and schizophre-
nia. Am J Med Genet B Neuropsychiatr Genet 135B: 94–101.
15. Addington AM, Gornick M, Duckworth J, Sporn A, Gogtay N, et al. (2005)
GAD1 (2q31.1), which encodes glutamic acid decarboxylase (GAD67), is
associated with childhood-onset schizophrenia and cortical gray matter volume
loss. Mol Psychiatry 10: 581–588.
16. Du J, Duan S, Wang H, Chen W, Zhao X, et al. (2008) Comprehensive analysis
of polymorphisms throughout GAD1 gene: a family-based association study in
schizophrenia. J Neural Transm 115: 513–519.
17. Straub RE, Lipska BK, Egan MF, Goldberg TE, Callicott JH, et al. (2007)
Allelic variation in GAD1 (GAD67) is associated with schizophrenia and
influences cortical function and gene expression. Mol Psychiatry 12: 854–869.
18. Zhao X, Qin S, Shi Y, Zhang A, Zhang J, et al. (2007) Systematic study of
association of four GABAergic genes: glutamic acid decarboxylase 1 gene,
glutamic acid decarboxylase 2 gene, GABA(B) receptor 1 gene and GABA(A)
receptor subunit beta2 gene, with schizophrenia using a universal DNA
microarray. Schizophr Res 93: 374–384.
19. Utge S, Soronen P, Partonen T, Loukola A, Kronholm E, et al. (2010) A
population-based association study of candidate genes for depression and sleep
disturbance. Am J Med Genet B Neuropsychiatr Genet 153B: 468–476.
20. Chambless DL, Caputo GC, Bright P, Gallagher R (1984) Assessment of fear of
fear in agoraphobics: the body sensations questionnaire and the agoraphobic
cognitions questionnaire. J Consult Clin Psychol 52: 1090–1097.
21. Sanacora G, Gueorguieva R, Epperson CN, Wu YT, Appel M, et al. (2004)
Subtype-specific alterations of gamma-aminobutyric acid and glutamate in
patients with major depression. Arch Gen Psychiatry 61: 705–713.
22. Jarius S, Stich O, Speck J, Rasiah C, Wildemann B, et al. (2010) Qualitative and
quantitative evidence of anti-glutamic acid decarboxylase-specific intrathecal
antibody synthesis in patients with stiff person syndrome. J Neuroimmunol.
23. Koerner C, Wieland B, Richter W, Meinck HM (2004) Stiff-person syndromes:
motor cortex hyperexcitability correlates with anti-GAD autoimmunity.
Neurology 62: 1357–1362.
24. Geis C, Weishaupt A, Grunewald B, Wultsch T, Reif A, et al. (2011) Human
stiff-person syndrome IgG induces anxious behavior in rats. PLoS One 6:
25. Donner J, Sipila T, Ripatti S, Kananen L, Chen X, et al. (2012) Support for
involvement of glutamate decarboxylase 1 and neuropeptide y in anxiety
susceptibility. Am J Med Genet B Neuropsychiatr Genet 159B: 316–327.
26. Marenco S, Savostyanova AA, van der Veen JW, Geramita M, Stern A, et al.
(2010) Genetic modulation of GABA levels in the anterior cingulate cortex by
GAD1 and COMT. Neuropsychopharmacology 35: 1708–1717.
27. Yoon JH, Maddock RJ, Rokem A, Silver MA, Minzenberg MJ, et al. (2010)
GABA concentration is reduced in visual cortex in schizophrenia and correlates
with orientation-specific surround suppression. J Neurosci 30: 3777–3781.
28. Lappalainen J, Sanacora G, Kranzler HR, Malison R, Hibbard ES, et al. (2004)
Mutation screen of the glutamate decarboxylase-67 gene and haplotype
association to unipolar depression. Am J Med Genet B Neuropsychiatr Genet
29. Ikeda M, Ozaki N, Yamanouchi Y, Suzuki T, Kitajima T, et al. (2007) No
association between the glutamate decarboxylase 67 gene (GAD1) and
schizophrenia in the Japanese population. Schizophr Res 91: 22–26.
30. Buttenschon HN, Lauritsen MB, El Daoud A, Hollegaard M, Jorgensen M, et
al. (2009) A population-based association study of glutamate decarboxylase 1 as
a candidate gene for autism. J Neural Transm 116: 381–388.
31. Rabionet R, Jaworski JM, Ashley-Koch AE, Martin ER, Sutcliffe JS, et al. (2004)
Analysis of the autism chromosome 2 linkage region: GAD1 and other candidate
genes. Neurosci Lett 372: 209–214.
32. Mosing MA, Gordon SD, Medland SE, Statham DJ, Nelson EC, et al. (2009)
Genetic and environmental influences on the co-morbidity between depression,
panic disorder, agoraphobia, and social phobia: a twin study. Depress Anxiety
33. Smoller JW, Block SR, Young MM (2009) Genetics of anxiety disorders: the
complex road from DSM to DNA. Depress Anxiety 26: 965–975.
34. Klauke B, Deckert J, Reif A, Pauli P, Domschke K (2010) Life events in panic
disorder-an update on ‘‘candidate stressors’’. Depress Anxiety.
35. Zhang TY, Hellstrom IC, Bagot RC, Wen X, Diorio J, et al. (2010) Maternal
care and DNA methylation of a glutamic Acid decarboxylase 1 promoter in rat
hippocampus. J Neurosci 30: 13130–13137.
36. Mannuzza S, Fyer AJ, Klein DF, Endicott J (1986) Schedule for Affective
Disorders and Schizophrenia–Lifetime Version modified for the study of anxiety
disorders (SADS-LA): rationale and conceptual development. J Psychiatr Res 20:
37. Robins LN, Wing J, Wittchen HU, Helzer JE, Babor TF, et al. (1988) The
Composite International Diagnostic Interview. An epidemiologic Instrument
suitable for use in conjunction with different diagnostic systems and in different
cultures. Arch Gen Psychiatry 45: 1069–1077.
38. Wittchen HU (1997) SKID-I: Strukturiertes Klinisches Interview fu ¨r DSM-IV,
Achse I: Psychische Sto ¨rungen. Goettingen: Hogrefe.
39. Gloster AT, Wittchen HU, Einsle F, Hofler M, Lang T, et al. (2009) Mechanism
of action in CBT (MAC): methods of a multi-center randomized controlled trial
in 369 patients with panic disorder and agoraphobia. Eur Arch Psychiatry Clin
Neurosci 259 Suppl 2: S155–166.
40. Reiss S, Peterson RA, Gursky DM, McNally RJ (1986) Anxiety sensitivity,
anxiety frequency and the prediction of fearfulness. Behav Res Ther 24: 1–8.
41. Frazer KA, Ballinger DG, Cox DR, Hinds DA, Stuve LL, et al. (2007) A second
generation human haplotype map of over 3.1 million SNPs. Nature 449:
42. Barrett JC, Fry B, Maller J, Daly MJ (2005) Haploview: analysis and
visualization of LD and haplotype maps. Bioinformatics 21: 263–265.
43. Menashe I, Rosenberg PS, Chen BE (2008) PGA: power calculator for case-
control genetic association analyses. BMC Genet 9: 36.
44. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, et al. (2007)
PLINK: a tool set for whole-genome association and population-based linkage
analyses. Am J Hum Genet 81: 559–575.
45. R Development Core Team (2010) R: A Language and Environment for
Statistical Computing. 2.11 ed. Vienna, Austria.
46. Coassin S, Brandstatter A, Kronenberg F (2010) Lost in the space of
bioinformatic tools: a constantly updated survival guide for genetic epidemiol-
ogy. The GenEpi Toolbox. Atherosclerosis 209: 321–335.
47. Cartharius K, Frech K, Grote K, Klocke B, Haltmeier M, et al. (2005)
MatInspector and beyond: promoter analysis based on transcription factor
binding sites. Bioinformatics 21: 2933–2942.
48. Lee PH, Shatkay H (2008) F-SNP: computationally predicted functional SNPs
for disease association studies. Nucleic Acids Res 36: D820–824.
49. Lee PH, Shatkay H (2009) An integrative scoring system for ranking SNPs by
their potential deleterious effects. Bioinformatics 25: 1048–1055.
Associations of GAD1 with Panic Disorder
PLoS ONE | www.plosone.org7 May 2012 | Volume 7 | Issue 5 | e37651