Lilly-Molecular Psychiatry Award Honorable Mention

Genetic investigation of chromosome 5q GABAA receptor subunit genes in schizophrenia


We previously performed a genome-wide linkage scan in Portuguese schizophrenia families that identified a risk locus on chromosome 5q31–q35. This finding was supported by meta-analysis of 20 other schizophrenia genome-wide scans that identified 5q23.2–q34 as the second most compelling susceptibility locus in the genome. In the present report, we took a two-stage candidate gene association approach to investigate a group of gamma-aminobutyric acid (GABA) A receptor subunit genes (GABRA1, GABRA6, GABRB2, GABRG2, and GABRP) within our linkage peak. These genes are plausible candidates based on prior evidence for GABA system involvement in schizophrenia. In the first stage, associations were detected in a Portuguese patient sample with single nucleotide polymorphisms (SNPs) and haplotypes in GABRA1 (P=0.00062–0.048), GABRP (P=0.0024–0.042), and GABRA6 (P=0.0065–0.0088). The GABRA1 and GABRP findings were replicated in the second stage in an independent German family-based sample (P=0.0015–0.043). Supportive evidence for association was also obtained for a previously reported GABRB2 risk haplotype. Exploratory analyses of the effects of associated GABRA1 haplotypes on transcript levels found altered expression of GABRA6 and coexpressed genes of GABRA1 and GABRB2. Comparison of transcript levels in schizophrenia patients and unaffected siblings found lower patient expression of GABRA6 and coexpressed genes of GABRA1. Interestingly, the GABRA1 coexpressed genes include synaptic and vesicle-associated genes previously found altered in schizophrenia prefrontal cortex. Taken together, these results support the involvement of the chromosome 5q GABAA receptor gene cluster in schizophrenia, and suggest that schizophrenia-associated haplotypes may alter expression of GABA-related genes.


Schizophrenia is a complex psychiatric disorder with a strong genetic component.1, 2 Nearly 40 independent genome-wide linkage scans of schizophrenia have been conducted to date.3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37 Meta-analysis of 20 of these scans identified chromosome 5q23.2–q34 from D5S2098 to D5S422 (123.8–162.1 Mb; National Center for Biotechnology Information (NCBI) Build 35 human genome assembly) as the second most significant risk locus in the genome.38 Subsequently, we reported linkage to an overlapping region at chromosome 5q31.1–q35.1 in a genome-wide scan of schizophrenia and schizoaffective depressed type (SA-D) families of Portuguese descent.32 The maximum linkage signal in 29 families was an NPL=3.09 (P=0.0012) at marker D5S820, falling just short of genome-wide significance based on simulations of the data.32 Higher density mapping in an expanded sample of 40 pedigrees produced a peak NPL=3.28 (P=0.00066) in the same region. Nominally significant linkage was obtained across a 35 cM region from D5S816 to D5S1456 located from 135.3 to 169 Mb. Although the 5q risk locus is large, the strength and consistency of the linkage findings warrants candidate gene association studies to identify the underlying schizophrenia risk gene.

The chromosome 5q risk locus contains a cluster of gamma-aminobutyric acid (GABA) A receptor subunit genes, GABRB2, GABRA6, GABRA1, and GABRG2, which are grouped within an 860-kb region at 5q34 that is 4 Mb telomeric to our peak at D5S820. A fifth GABAA receptor subunit gene, GABRP, is located 12 Mb telomeric to our peak at the boundary of our linkage region, which nevertheless could be within range to produce this linkage signal. As reviewed by Steiger and Russek,39 the GABAA receptor is a neuronal ligand-gated ion channel that is the major inhibitory receptor in the adult brain via its function in triggering chloride ion influx, resulting in cell hyperpolarization. Numerous receptor subunits have been identified and each displays different temporal and spatial patterns of expression. The subunits assemble into pentamers, typically of two alpha, two beta, and one gamma subunit. The most widely expressed GABAA receptor in the adult brain is composed of alpha 1, beta 2, and gamma 2 subunits, encoded by GABRA1, GABRB2, and GABRG2, respectively. These three genes are all located in the 5q GABAA receptor subunit gene cluster. The cluster also contains GABRA6 that encodes the alpha 6 subunit expressed in granule cells of the cerebellum and the cochlea.40 The pi subunit encoded by GABRP, which influences drug effects on the receptor,41 has been detected in the hippocampus and temporal cortex, as well as several peripheral tissues.42

The 5q GABA genes are plausible candidates for the schizophrenia risk gene in the 5q locus based on a converging body of literature implicating GABA system dysfunction in schizophrenia pathophysiology.43, 44, 45, 46, 47, 48, 49 In schizophrenia patient brain samples, markers of GABAergic neurons are reduced in expression and postsynaptic receptor complexes are increased in number.47, 50, 51 Furthermore, there is evidence that neuregulin 1 (NRG1) and dysbindin (DTNBP1), two schizophrenia candidate risk genes,33, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65 interact with the GABAA receptor. For example, treatment of the rat hippocampus with NRG1 resulted in lower expression of GABAA receptor subunit transcripts,66 while dysbindin likely colocalizes with GABAA receptor subunits in the hippocampus, cortex, and cerebellum through its association with beta-dystrobrevin in the dystrophin protein complex.67, 68, 69 Despite the evidence for GABA system alterations in schizophrenia patients, there has been little investigation of association between GABA receptor genes and schizophrenia, although association with the GABRB2 gene was recently reported in a Chinese patient sample.70

We performed an association study of the 5q GABAA receptor subunit gene cluster in schizophrenia based on: (1) strong meta-analytic evidence for linkage between schizophrenia and chromosome 5q, (2) linkage to 5q in our Portuguese schizophrenia families, (3) evidence for involvement of the GABA system in schizophrenia, and (4) location of the GABAA receptor subunit gene cluster in the 5q risk locus. To increase the chance of detecting association, we screened a Portuguese patient sample that was ascertained in the same manner as the Portuguese family sample, which demonstrated linkage to 5q. Positive associations were followed up in an independent sample of German patients. We subsequently explored whether GABAA receptor subunit gene haplotypes that were found to be associated with schizophrenia were related to altered transcript levels of the GABAA genes and groups of GABA-related genes. We also tested for disease-related expression differences in schizophrenia patients compared to their unaffected siblings. Our findings provide evidence for associations between schizophrenia and the GABRA1, GABRP, and GABRB2 genes, and suggest relationships between disease-associated haplotypes and transcript expression, as well as disease-specific transcript alterations.

Materials and methods


Characteristics of the samples utilized in the current study are provided in Table 1.

Table 1 Characteristics of samples used in association and microarray expression analyses

Portuguese-descent association and microarray mRNA expression samples

The subjects utilized in our association analyses and microarray expression analyses originated from continental Portugal and the islands of the Azores and Madeira, and have been previously described.32 Unrelated patients and control individuals were also ascertained from the Azorean immigrant population of Fall River, USA, and were confirmed to have all four grandparents from the Azores, Madeira, or continental Portugal. Controls consisted of unrelated individuals ascertained through a brochure left in community facilities (eg churches, community centers, doctor's offices, large employers) and four unaffected individuals that married into our linkage pedigrees.32 Best estimate diagnoses of schizophrenia or SA-D according to DSM-IV criteria were made by two blinded researchers after review of clinical information, Diagnostic Interview for Genetic Studies (DIGS),71 Operational Criteria Checklist for Psychotic Illness (OPCRIT),72 and written narratives. The study received approval by all appropriate Institutional Review Boards and subjects provided informed consent.

Our first-stage association sample consisted of Portuguese-descent parent–proband trios (N=111; 30% continental Portugal, 55% Azores, 15% Madiera), unrelated cases (N=321; 38% continental Portugal, 45% Azores, 17% Madiera), and control individuals (N=242; 12% continental Portugal, 85% Azores, 3% Madiera). All trio probands and cases were diagnosed with schizophrenia. In all, 12 parent–proband trios and six cases were also present in our previously reported linkage pedigrees.32 We used the Genetic Power Calculator73 to determine that the trios sample had 28 and 44% power, and the case–control sample had 58 and 81% power, to detect a risk allele with 10% frequency and 20% frequency, respectively, with a multiplicative genotype relative risk of 1.5. Given the modest power of the separate samples, association results for the parent–proband trios and for the unrelated cases and controls were combined (described below), as the collective sample had 75 and 92% power under the same assumptions. Tests for stratification74 using 60% of the collective sample (94 trios, 152 cases, and 149 controls) and 46 unlinked markers did not detect population substructure between the trios and case–control samples, nor between the cases and controls (75 and data not shown).

The Portuguese microarray mRNA expression sample has been described previously.59 Briefly, the sample consisted of primary leukocyte cell preparations from 33 age- and gender-matched sibling pairs (22 female, 11 male pairs; 32 schizophrenia and one SA-D discordant pairs) selected from our linkage pedigrees.32 Of the 33 sibpairs, 32 sibpairs were also genotyped for the same set of markers as the association sample. Of these individuals, 17 were cases in our association sample.

German association sample

Our second-stage association sample consisted of parent–proband trios (N=238) from Germany and Hungary, and have been previously described.76 Probands met Research Diagnostic Criteria77 for schizophrenia (N=228) or SA-D (N=10). The study received approval by all appropriate Institutional Review Boards and subjects provided informed consent.


Single nucleotide polymorphisms (SNPs) spanning the 5q GABAA receptor subunit genes were selected from the NCBI dbSNP database (, the Celera Discovery System database (, and the International HapMap Project ( We also genotyped five GABRB2 SNPs previously reported to be associated with schizophrenia in Chinese patients70 (rs187269, rs252944, rs194072, rs1816071, and rs6556547; note that rs6556547 in the current study is B217G1584 in the previous report). SNP genomic positions were according to the NCBI Build 35 human genome assembly ( Genotyping was performed by mass spectrometry as described previously78 using amplification and extension primers designed by SpectroDesigner software (Sequenom, La Jolla, CA, USA). Only genotype data for those SNPs that met the following quality control standards were further analyzed: (1) >85% of attempted genotypes were successful (mean=95.7% in Portuguese sample; mean=97.8% in German sample), (2) parental alleles were in Hardy–Weinberg equilibrium, (3) <1% of chromosomes transmitted from parents to probands had Mendelian inheritance errors, (4) >2% minor allele frequency. Marker rs1037715 was genotyped in triplicate in the Portuguese sample using different amplification and extension primers for each assay, and the concordance rate between assays was 98.5%. SNP primer sequences are available in Supplementary Table 1.

Linkage disequilibrium and association analysis methods

The Haploview program79 was used to determine linkage disequilibrium (LD) between pairs of markers by calculation of the D′ statistic,80 which ranges from D′=0 (complete recombination) to D′=1.0 (no recombination). Haplotype blocks (regions of minimal historical recombination that contain only a few common haplotypes) were defined as previously described.81 LD between haplotype blocks was determined by the Haploview program, which considers haplotypes within blocks as alleles of a multiallelic marker using Hedrick's multiallelic extension to the D′ statistic.82 Maximum-likelihood expectation (MLE) haplotypes within each haplotype block were reconstructed using an expectation maximization algorithm.83

Association analyses were performed using a narrow phenotype that included schizophrenia and SA-D, which we refer to collectively as ‘schizophrenia’. Single markers and MLE haplotypes were tested for association with schizophrenia by the transmission disequilibrium test (TDT)84 for parent–proband trios, and by χ2 test for the unrelated cases and controls. Note that the trio probands were not used as cases in the case–control analyses, thus the trios and case–control samples did not overlap. Since the power of each sample was not high to detect associations with lower frequency alleles, the results for the trios sample and the case–control sample were combined. This was carried out by reformatting the TDT results and χ2 results as a mean, observed, and variance on the number of risk alleles transmitted in the trios or found in the unrelated cases. A combined standard normal Z-score was calculated, where the sum of the deviations from expected frequencies (under the null hypothesis of no association) of each sample was divided by the sum of the variances of each sample, as previously described.59, 85 Nominal significance levels were obtained from the Z distribution. This approach provided SNP allele- or haplotype-specific nominal significance levels. Odds ratios (OR) were computed using the method of Morris and Gardner,86 with the logit method used to calculate the OR of the combined Portuguese trios and case–control sample.

Associations that surpassed a nominal two-tailed P<0.05 in our Portuguese sample were subsequently examined in an independent German sample. Single markers and MLE haplotypes were tested for association to schizophrenia by the TDT.84 One-tailed P-values, indicated by (*), are reported for the German sample for all alleles and haplotypes that were associated in the same manner as in the Portuguese sample (eg, higher frequency in Portuguese patients and overtransmitted to German trio probands). We considered an allele or haplotype to be replicated in the German sample only if the association was in the same manner with a one-tailed P<0.05*. This two-stage approach has the advantage that it is not overly punitive in the first stage, but relies on replication in a second sample for support of association. Relative risks were computed as the ratio of the number of transmissions to the number of nontransmissions, as determined by the TDT. One-tailed P-values are also reported for the Portuguese and German samples for GABRB2 alleles and haplotypes that supported schizophrenia association results previously reported in a Chinese study.70

Microarray mRNA expression methods

As described previously,59 total RNA was extracted from blood leukocyte cell preparations from discordant sibpairs selected from our Portuguese linkage pedigrees.32 Microarray samples were labeled and processed according to standard protocols, hybridized to the Human Genome U133A GeneChip® (Affymetrix), washed and stained on the Fluidics Station (Affymetrix) according to the EukGE-WS2 protocol, and scanned using the Agilent G2500A Gene Array Scanner. Scans were subjected to a priori quality control criteria and matching criteria to obtain 33 high-quality matched sibpairs for expression analyses. The scan files were normalized to a median intensity value of 500 using Affymetrix MicroArray Suite 5.0. Normalized expression data for the 5q GABAA receptor subunit genes and GABA gene sets reported in this study are available in Supplementary Table 2.

Validation of microarray data by real-time quantitative PCR

Microarray expression levels of the discordant sibpair expression sample used in the current study have previously been successfully validated for five transcripts of interest and two reference genes using real-time quantitative RT-PCR.59, 87 In those studies, the fold changes reported by the microarray and RT-PCR methods for the seven genes were highly correlated (r=0.95) based on triplicate assays for 19 sibpairs that had sufficient RNA for both procedures. In light of this previous work, and due to the limited amount of RNA available for validation in only a subset of the sample, we did not perform additional validation efforts in the present study.

Microarray expression data analysis

Linear regression analysis of GABA transcript expression

We tested whether two GABRA1 haplotypes, Hap10A and Hap10B, that had evidence for association with schizophrenia in both our Portuguese and German association samples had an effect on transcript expression levels of the 5q GABAA receptor subunit genes. Haplotypes were determined for each individual from 32 of the 33 discordant schizophrenia sibpairs for which we had expression data. The observed frequencies of the tested haplotypes were as follows: GABRA1 Hap10A: 17% +/+, 52% +/−, 31% −/−; and GABRA1 Hap10B: 0% +/+, 37% +/−, 63% −/−. GABRP Hap18D, which also had evidence for association in both patient samples, was not tested because the haplotype frequency was too low to be informative for these analyses. Expression data were linearly regressed on genotypes regardless of schizophrenia diagnosis (ie, the effect of a particular haplotype was investigated in the schizophrenia and unaffected siblings together). We only tested for haplotype effects on expression of the five 5q GABAA receptor subunit genes, GABRA1, GABRA6, GABRB2, GABRG2, and GABRP. We utilized the standard biometrical model88 with additive effect only. For GABRA1 Hap10A, we also tested an additive+dominant model, since all three genetic classes were observed in the sample. Thus, a total of 15 tests were performed (five transcripts × three genetic models (Hap10A: two models+Hap10B: one model)). Significance of each model was assessed by F-test, and the additive and additive+dominant models for Hap10A were compared by ANOVA to determine if there was a significant dominance component. Nominal P-values are reported.

Gene Set Enrichment Analysis of GABA gene sets

We used the Gene Set Enrichment Analysis (GSEA) method89 to test whether two haplotypes that were associated with schizophrenia (GABRA1 Hap10A and GABRA1 Hap10B) were also associated with coordinate alterations in expression of related genes. Details of the analysis are provided in Supplementary Table 3. Briefly, for a given pairwise comparison (eg Hap10A+/+ vs Hap10A−/−), we determined whether the expression of a related set of genes (the ‘gene set’) was coordinately altered (ie, either up- or down-regulated) in one haplotype group compared to the other. To determine nominal significance levels adjusted for the number of genes on the microarray included in the analyses, the category labels of the samples (eg Hap10A+/+ or Hap10A−/−) were randomly permuted 10 000 times, and the observed result was compared with the permutation results. To interpret the results with respect to the number of gene sets and haplotype comparisons tested, we calculated a false discovery rate (FDR),90 which is the expected proportion of false results (Type I errors) among the results that surpassed a particular significance threshold. The FDR was calculated as nP/R, where n is the number of independent tests, P is the significance threshold applied to the results, and R is the number of reported results that met or surpassed P.

Gene sets were created consisting of coexpressed genes of each of the 5q GABAA receptor subunit genes. For each gene (GABRA1, GABRA6, GABRB2, GABRG2, and GABRP), we derived four gene sets comprised of: (a) the 50 nearest expression neighbors of the gene in the Genomics Institute of the Novartis Research Foundation (GNF) expression atlas version 191 (consisting of 12 000 genes), using either all of the tissues or only the brain tissues in the atlas, and (b) the 100 nearest expression neighbors of the gene in the GNF expression atlas version 291, 92 (consisting of 22 000 genes), using either all of the tissues or only the brain tissues in the atlas. This resulted in a total of 20 GABA gene sets. The gene sets were named according to the GABAA receptor subunit gene, the atlas version, and the atlas tissues used (eg GABRA1-v1-all tissues).

Peripheral leukocyte microarray samples from the 32 discordant sibpairs were grouped to reflect as closely as possible, given the restriction to two-class comparisons in the GSEA method, three possible genetic effects of the associated haplotypes: Hap+/+ vs −/− (similar to an additive model but excluding Hap+/− individuals), Hap+/+ vs +/− and −/− (recessive model), or Hap+/+ and +/− vs −/− (dominant model). Note that only the Hap +/− vs Hap−/− model could be tested for GABRA1 Hap10B, since the corresponding Hap+/+ homozygous classes were not observed in the sample. We performed a total of 80 tests for haplotype effects on gene set expression (20 GABA gene sets × four genetic models (Hap10A: three models+Hap10B: one model)). GSEA was also performed to compare the schizophrenia and unaffected siblings (regardless of the presence of particular haplotypes) for differences in expression of the 20 GABA gene sets.

Analysis of schizophrenia and unaffected sibling pairs

Scheffe's tests were performed to identify expression differences between schizophrenia and unaffected siblings in transcripts of the 5q GABAA receptor subunit genes. A paired design was utilized in which expression differences within each sibpair were calculated and subsequently averaged across all sibpairs.


Association between schizophrenia and 5q GABAA receptor subunit genes

To test for association with schizophrenia, 132 SNPs spanning the coding regions and 10 kb of flanking sequence of GABRB2, GABRA6, GABRA1, GABRG2, and GABRP were genotyped in a Portuguese sample consisting of parent–proband trios (N=111), unrelated cases (N=321), and controls (N=242). Figure 1 demonstrates the chromosomal order and distances between the genes, with GABRB2, GABRA6, GABRA1, and GABRG2 (listed centromeric to telomeric) clustered within an 860-kb region, and GABRP located 8 Mb telomeric to the cluster. Figure 2 displays the SNP positions relative to the five genes. Significant associations were detected between schizophrenia and 19 SNPs (P<0.05) spanning the gene cluster (Table 2). The majority of associated SNPs were in GABRA1, with some in GABRB2, GABRA6, and GABRP. Association results for all 132 SNPs spanning the cluster are provided in Supplementary Table 4.

Figure 1

Genomic structure of the chromosome 5q GABAA receptor subunit genes tested for association with schizophrenia. Orientation of genes is indicated by direction of arrows. Physical distances between genes are shown.

Figure 2

LD and haplotype block structure of the 5q GABAA receptor subunit genes. Gene structure is shown, with vertical lines indicating exons, above the bar representing the chromosome. Markers are displayed relative to gene location. LD structure between marker pairs is indicated by the colored matrices. Haplotype blocks spanning the GABAA receptor subunit genes are shown as yellow rectangles. Genomic positions are according to the NCBI Build 35 human genome assembly. Figure generated using LocusView version 2.0 (Petryshen and Kirby, unpublished software;

Table 2 SNP and haplotype associations between schizophrenia and 5q GABAA receptor subunit genes

To further investigate these associations, we performed haplotype-based analyses. We determined the pairwise LD between all SNPs and identified 21 haplotype blocks of strong LD spanning the gene cluster (Figure 2). The haplotype blocks averaged 25 kb in size and consisted of two to eight common haplotypes with frequencies >2%. Haplotype alleles and population frequencies are presented in Supplementary Table 5.

As shown in Table 2, significant associations (P<0.05) were detected with several haplotypes in the gene cluster. In GABRA1, associations were found with four haplotypes in adjacent haplotype blocks 9 and 10 that span the gene and contain all of the GABRA1 SNPs that were individually associated with schizophrenia. Haplotypes Hap9A in block 9 and Hap10A in block 10 were in excess in patients (parent–proband transmissions and unrelated cases) compared to controls (parent–proband nontransmissions and unrelated controls) (both P=0.0019). Hap9C and Hap10B were at lower frequencies in patients (P=0.0013 and 0.027, respectively). Since LD was strong between haplotype blocks 9 and 10 (D′=0.9), we tested haplotypes spanning both blocks for association. Hap9A+10A was at higher frequency in patients (P=0.00086) and Hap9C+10B was at lower frequency (P=0.00062). Associations were also detected with two other 5q GABAA receptor subunit genes. In GABRA6, Hap8C was under-represented (P=0.0067) in patients compared to controls. In GABRP, a low-frequency (<5%) haplotype, Hap18D, was in excess in patients (P=0.0024). Adjacent GABRP haplotypes Hap20A and Hap21B were less frequent in patients (P=0.042 and 0.038, respectively). As shown in Table 2, comparison of the number of transmissions and nontransmissions in the trios sample, and the frequencies in the case–control sample, found that the haplotype associations were not due to one sample alone, but rather the effects were consistent in the two samples. Frequencies of the above haplotypes were similar in a subset of 31 trios and 59 unrelated cases that had a positive family history, defined as having at least one first- or second-degree relative with schizophrenia (data not shown).

Replication of association between schizophrenia and 5q GABAA receptor subunit genes

We sought to replicate the nominally significant Portuguese SNP and haplotype associations in an independent sample of German parent–proband trios (N=238). As shown in Table 2, associations were replicated at P<0.05 for three of the 19 SNPs tested: rs168697 in GABRB2 (P=0.010*), rs1037715 in GABRA1 (P=0.0015*) and rs7736504 in GABRP (P=0.023*). Note that these two SNPs were associated in the same manner as the Portuguese sample, in that the alleles found in excess in the Portuguese patients were also overtransmitted to the German probands. We reconstructed haplotypes in the German sample using the same haplotype block structure defined in the Portuguese sample, since the LD structures of the two samples were for the most part identical (data not shown). Of the 10 haplotypes tested, associations were replicated at P<0.05 for four haplotypes in GABRA1 and GABRP (Table 2). In GABRA1, risk haplotypes Hap10A and Hap9A+10A were overtransmitted to affected probands (P=0.043* and P=0.023*, respectively), while Hap10B was undertransmitted to probands (P=0.0036*). Note that Hap10B is uniquely tagged by the T allele of rs1037715(C/T), one of the two associated SNPs in this sample, in that the T allele was found only on Hap10B and no other haplotypes in block 10 (Supplementary Table 5). In GABRP, risk haplotype Hap18D was overtransmitted to probands (P=0.023*; Table 2). This haplotype is uniquely tagged by the A allele of rs7736504(G/A), one of the two SNPs associated in this sample. After stringent Bonferroni correction for the 29 tests (19 SNPs and 10 haplotypes) performed in the German sample, only the association with rs1037715 in GABRA1 remained significant at P=0.044* (nominal P=0.0015* × 29).

Support for association between schizophrenia and GABRB2

We specifically investigated GABRB2 SNPs previously reported to be associated in Chinese schizophrenia patients.70 None of the SNPs (rs187269, rs252944, rs194072, rs1816071, and rs6556547) were significantly associated in either the Portuguese or German patient samples (all P>0.05; Supplementary Table 4). The alleles of each of the five SNPS that were at higher frequency in the Chinese cases together comprised a haplotype, Hap3C, found at 8% frequency in our samples. This haplotype was overtransmitted to probands in the German sample (P=0.035*; Table 2), and thereby provided support for the association reported in the previous study.70 Hap3C was not associated in the Portuguese parent–proband trios and case–control sample (P=0.92; Table 2). Post hoc analyses of the Portuguese trios and case–control samples separately found that Hap3C was overtransmitted to trio probands (20 transmissions vs 11 nontransmissions; P=0.053*), but the cases and controls had similar haplotype frequencies (6.3% cases vs 8.1% controls; P=0.24).

Effects of haplotypes associated with schizophrenia on GABAA receptor subunit expression levels

Disease-associated haplotypes may exert their effect by altering their own expression or the expression of other genes. Thus, we examined whether the haplotypes that we found associated with schizophrenia in both the Portuguese and German samples were related to alterations in transcript levels of the 5q GABAA receptor subunit genes in a sample of 32 discordant sibling pairs. We determined the genotype of each individual for haplotypes GABRA1 Hap10A and GABRA1 Hap10B. We did not test GABRP Hap18D, even though we detected association with this haplotype in both the Portuguese and German samples, because it was too rare to be informative. Haplotypes from all individuals (both affected and unaffected) were analyzed together in order to maximize statistical power. Microarray mRNA expression levels of each of the five 5q GABAA receptor subunit genes were linearly regressed against genotypes of GABRA1 Hap10A or GABRA1 Hap10B. A significant effect of GABRA1 Hap10A on lower GABRA6 expression was found under a dominant model (P=0.016, 1 df; Table 3), with Hap10A+/+ and +/− individuals having 1.7-fold lower expression of GABRA6 compared to Hap10A−/− individuals.

Table 3 Altered expression of GABAA receptor subunit genes and GABA gene sets in peripheral leukocytes from schizophrenia discordant sibpairs

Exploration of effects of haplotypes associated with schizophrenia on GABA gene set expression

Changes at the transcript or protein level of a GABAA receptor subunit could affect the physiologic state of the cell by altering cell electrical potential and downstream signaling molecules. Thus, we explored whether the associated GABRA1 haplotypes had effects on expression of groups of genes related to the GABAA receptor subunit genes. Since functionally related genes are often coregulated,93, 94 we created sets of genes based on coexpression with each GABAA receptor subunit gene. Specifically, for each of the five GABAA receptor subunit genes (GABRA1, GABRA6, GABRB2, GABRG2, and GABRP), we created four gene sets using either of two public expression atlases (referred to as ‘v1’ and ‘v2’)91, 92, and either all tissues or only the brain tissues in each atlas (referred to as ‘all tissues’ and ‘brain’), resulting in a total of 20 gene sets. The 32 discordant sibpairs described above were classified based on presence of a particular haplotype into two categories that reflected additive, dominant, and recessive genetic models (as detailed in the Materials and methods section). For each haplotype–gene set analysis, we used GSEA89 to examine the microarray mRNA expression data for the presence of gene set members among those genes that were the most differentially expressed between the two haplotype categories compared. Of the total 80 haplotype-gene set comparisons tested, two results were significant at a threshold of P0.01. This corresponds to a 40% FDR, thus one of the two results may be false.

Both of the two significant haplotype–gene set results involved GABRA1 Hap10B, which we found under-represented in schizophrenia patients. As reported in Table 3, Hap10B was related to higher expression of two GABA gene sets: Hap10B+/− individuals had 1.21-fold higher expression of the GABRA1-v1-all tissues gene set, and 1.25-fold higher expression of the GABRB2-v2-all tissues gene set, compared to Hap10B−/− individuals (P=0.0100 and 0.0014, respectively). Note that the genes in these two gene sets do not overlap, and thus the effects of Hap10B on the gene sets were not due to shared genes. As shown in Table 3, the GABRA1 gene set primarily contains presynaptic and vesicle-associated proteins, and the GABRB2 gene set contains many neurotransmitter receptors and other receptors. Post hoc analyses of the individual genes in the two gene sets determined that many genes had significant expression differences on their own between the particular haplotype groups (uncorrected Student's t-test P0.0032; Supplementary Table 3). Supplementary Table 3 lists the gene set members and their expression fold changes between Hap10B+/− and Hap10B−/− individuals.

Transcript expression alterations in schizophrenia patient peripheral leukocytes

Previous studies of postmortem prefrontal cortex from schizophrenia patients have reported alterations in some of the 5q GABAA receptor subunit genes and in GABAergic genes.95, 96, 97 Therefore, we investigated the discordant sibling pairs for expression differences between the schizophrenia siblings and the unaffected siblings (ie, we tested for an effect of disease, rather than effects of haplotypes as reported above). Linear regression analysis of each of the five GABAA receptor subunit genes detected an average 1.37-fold decrease in GABRA6 transcript expression in the schizophrenia siblings (P=0.027; Table 3). Analysis of the 20 GABA gene sets using GSEA detected an average 1.14-fold decrease of the GABRA1-v1-all tissues gene set in schizophrenia siblings (P=0.025; FDR=50%). We have also found that this gene set has lower expression in the prefrontal cortex of schizophrenia patients compared to controls (P=0.005; unpublished data) in a postmortem sample obtained from the Stanley Medical Research Foundation that has been described previously.98 Supplementary Table 3 lists the GABRA1 gene set members and their expression fold changes between schizophrenia patients and unaffected individuals.


Numerous pharmacological, imaging, and postmortem studies have implicated the GABA system in schizophrenia pathophysiology.47, 48, 49 The current study has presented several lines of evidence for the specific involvement of GABAA receptor subunit genes located in the 5q schizophrenia risk locus. We detected association between schizophrenia and both GABRA1 and GABRP in a patient sample of Portuguese descent. Importantly, we subsequently replicated the GABRA1 and GABRP associations in an independent sample of German origin. We also obtained supportive evidence for association with GABRB2 that was previously reported in a Chinese patient sample.70 Furthermore, exploratory analyses suggested that the associated GABRA1 haplotypes had effects on mRNA expression levels of some of the 5q GABAA receptor genes and related groups of genes, including some synaptic and vesicle-associated genes that have been reported to be altered in schizophrenia prefrontal cortex.96

We detected associations between two GABRA1 haplotypes and schizophrenia in our Portuguese sample that we subsequently replicated in our German sample. A ‘risk’ haplotype at 50% population frequency, Hap10A, was in excess in the Portuguese patients and was overtransmitted to German probands. In contrast, a ‘protective’ haplotype at 15% frequency that spanned the same region, Hap10B, was under-represented in Portuguese patients and undertransmitted to German probands. These haplotypes contain the SNP that had the strongest association in our second-stage German sample, which remained significant at P<0.05 after stringent Bonferroni correction for the number of tests performed in this sample. It will be important to determine whether only one of these haplotypes accounts for the observed association; however, to do this effectively requires a much larger sample than we currently have. We also detected associations in both of our patient samples with a rare risk haplotype in GABRP. None of the SNP alleles that uniquely mark these GABRA1 and GABRP haplotypes are known to be functional, and are thus not likely themselves directly responsible for the altered risk of schizophrenia observed, but are likely in LD with other functional or regulatory alleles. Identifying the causal alleles will require further sequencing of these genes to assemble a comprehensive set of sequence variants, followed by genotyping in large, well-characterized populations of schizophrenia patients.

We specifically investigated a GABRB2 risk haplotype containing SNP alleles reported to be in excess in Chinese schizophrenia patients,70 and obtained modest support for association in our German and Portuguese parent–proband trios samples. The observed association in our trios samples indicates that the Chinese report was unlikely due to population stratification between cases and controls, and likewise our trios results are not likely due to transmission distortion. Association was not detected with this haplotype in our Portuguese case–control sample, however. The reason is unclear, and may only become apparent after additional studies in independent samples are carried out to delineate the role of this haplotype in schizophrenia risk. Evidence has also been presented for association with GABRG2 in Finnish patients;99 however, the data have not yet been published to permit thorough comparison with our findings. Other 5q GABAA receptor subunit gene haplotypes were also associated in our Portuguese sample but were not replicated in the German sample, suggesting that the Portuguese associations may be Type I errors. However, investigation of these haplotypes in other patient samples may be necessary to conclusively rule out their involvement in schizophrenia.

Our results and those of others70, 99 suggest that the functionally related GABRA1, GABRB2, GABRG2, and GABRP genes on chromosome 5q may all confer risk of schizophrenia. This finding is not overly surprising since the first three genes encode subunits that associate together in the major GABAA receptor in the adult brain,100, 101 and the latter gene encodes a regulatory subunit that modulates the effects of drugs on the receptor.41 Given that these subunits associate together in a receptor, epistatic interactions between subunit genes are possible. However, investigation of interactions will require larger samples than we currently have in order to achieve adequate statistical power, particularly for less frequent haplotypes (eg our Portuguese sample had <75% power to detect interactions between the GABRA1 and GABRB2 haplotypes).

Gene polymorphisms that are involved in disease susceptibility may function by altering the encoded polypeptide sequence, transcript expression levels, or temporal expression patterns, for example. Expression alterations may occur either by a sequence variant in a gene dictating the level of transcription of the gene itself, or the processing or stability of its mRNA transcript. Alternatively, a gene sequence variant may alter the expression of other genes. Changes in the expression level or encoded protein of a GABAA receptor subunit could affect the physiologic state of the cell through alterations in cell electrical potential and downstream signaling molecules, since the GABAA receptors flux chloride. Such changes could affect regulation of a multitude of genes through various signal transduction pathways.

Our exploratory analyses of the effects of GABRA1 haplotypes that we found associated with schizophrenia on expression of the GABA genes and GABA-related gene sets produced several interesting findings. We did not detect significant effects of the haplotypes on expression of their own transcripts, although our sample may have insufficient power to detect such effects, particularly for less frequent haplotypes. However, we did observe potential effects of the haplotypes on expression levels of other 5q GABA subunit genes and coexpressed genes. Specifically, the GABRA1 Hap10A risk haplotype was correlated with lower expression of the GABRA6 gene, and the GABRA1 Hap10B protective haplotype was related to coordinate expression alterations in gene sets that are coexpressed with GABRA1 and GABRB2. Furthermore, many of the genes in these two gene sets on their own had nominally significant expression differences between the haplotype groups. These preliminary findings suggest that one or more sequence variants residing on haplotypes of GABRA1 function to regulate biological pathways in which itself and the GABRB2 gene, which we and others70 have implicated in schizophrenia risk, may operate. However, determination of the exact nature of the relationship between these haplotypes, the altered biological pathways, and the pathophysiological findings observed in the brains of schizophrenia patients requires further investigation. Interestingly, the altered GABRA1 gene set primarily consist of synaptic and vesicle-associated proteins (eg dynamin, N-ethylmalemide-sensitive factor (NSF), synaptotagmin, synaptogyrin, and synaptophysin), many of which have previously been reported to be downregulated in schizophrenia prefrontal cortex.96 The altered GABRB2 gene set contains a more diverse set of primarily membrane-bound and extracellular proteins, including neurotransmitter receptors (eg 5-hydroxytryptamine receptor 1F, nicotinic acetylcholine receptor beta 3, GABA receptor rho 2 subunit, GABA receptor theta subunit, and N-methyl-D-aspartate receptor 2A subunit (GRIN2A)) and other receptors (eg retinoic acid receptor beta, retinoid X receptor gamma, estrogen-related receptor alpha, and fibroblast growth factor receptor 2).

Lower expression of presynaptic secretory machinery has been postulated as a specific schizophrenia-related defect.102 Our results demonstrated that a GABRA1 gene set which, as described above, contains many synaptic and vesicle-associated genes, was coordinately decreased an average of 1.14-fold in schizophrenia siblings compared to unaffected siblings. Further support for this defect has been found in our unpublished observations, in which this gene set has lower expression in the prefrontal cortex of schizophrenia patients compared to controls in a postmortem sample obtained from the Stanley Medical Research Foundation.

We point out two important considerations in interpreting our mRNA expression results. The first is that we applied nominal significance thresholds to our gene set analysis results that were adjusted for the large number of genes on the microarray, but were not corrected for the number of comparisons performed. Therefore, to interpret our results in light of the multiple tests, we determined the FDR for our analyses. The FDR's were 40 and 50% for our haplotype-based and disease-based gene set analyses, respectively, indicating that one of the two significant gene sets in the haplotype-based analyses, and the only significant gene set in the disease-based analyses, may be false positives. The second consideration is that, as we have discussed previously,59 we utilized peripheral leukocytes isolated from blood, rather than postmortem brain tissue. Our rationale is that artefacts introduced by postmortem hypoxia, anatomical inconsistencies, and postmortem interval in postmortem tissue studies are not an issue with leukocyte samples. However, it is unknown the extent to which this cell population reflects transcript expression profiles of the brain. Nonetheless, many brain-related genes and metabolic transcripts that have altered expression in schizophrenic postmortem prefrontal cortex are also altered in peripheral tissues, including lymphocytes,103, 104, 105 arguing for the validity of using peripheral leukocytes to measure schizophrenia gene expression alterations. Of relevance to the current study, murine T lymphocytes express functional GABAA receptors that are pharmacologically similar to their central nervous system counterparts,106 and GABA receptors have been found in pancreatic islets, the gastrointestinal tract, ovaries, and the adrenal medulla.107 Furthermore, we detected significant haplotype effects on gene sets containing synaptic and vesicle-associated genes that have previously been implicated in schizophrenia,96 indicating that genes that are typically considered brain-specific are in fact expressed in peripheral leukocytes. Clearly, the next step will be to determine in a large postmortem brain sample whether the haplotypes also affect these GABA-related pathways in the brain.

Interestingly, a recent study found that treatment of the rat hippocampus with NRG1 resulted in significantly lower mRNA expression of the GABAA receptor alpha 1 and alpha 2 subunits, and modestly decreased GABAA receptor beta 2 and gamma 2 subunit expression.66 Furthermore, different isoforms of NRG1 appear to act as attractants for migrating cortical GABAergic interneurons.108 These findings are highly intriguing given the reported association between the NRG1 gene and schizophrenia,33, 52, 53, 54, 55, 56, 57, 58, 59 and suggest that defects in GABAA receptor regulation by NRG1 is a possible mechanism by which NRG1 contributes to schizophrenia. Indeed, in a previous study using the same 33 discordant sibpairs as in the present study, we found higher expression of the SMDF isoform of NRG1 in subjects with schizophrenia.59 Thus, the conferred risk of schizophrenia by NRG1 and the 5q GABAA receptor subunit genes may be through a common GABA-related pathway, lending support to the GABA hypothesis of schizophrenia pathophysiology.

In conclusion, the current study has obtained evidence from multiple fronts that the GABAA receptor subunit genes located on chromosome 5q are involved in schizophrenia risk and pathophysiology. We have presented evidence for replicated associations between two GABAA receptor subunit genes and schizophrenia in two European samples, the most compelling results being for GABRA1. Our study also provides modest support for a previously reported GABRB2 association. Furthermore, we found relationships between associated GABRA1 haplotypes and expression levels of GABA genes, as well as gene sets that may be coregulated with the GABA genes. Taken together with historical and more recent data implicating the GABA system in schizophrenia, our data suggest that further investigations of these genes are warranted to understand the complex role that the GABA subunits and related genes play in schizophrenia pathogenesis.


  1. 1

    Kendler KS . Schizophrenia genetics. In: Sadock J, Sadock VA (eds). Kaplan and Sadock's Comprehensive Textbook of Psychiatry. Lippincott, Williams and Wilkins: Philadelphia, 2000.

  2. 2

    Owen MJ, O'Donovan MC, Gottesman II . Schizophrenia. In: McGuffin P, Owen M, Gottesman II (eds) Psychiatric Genetics and Genomics. Oxford University Press: Oxford, 2002, pp 247–256.

  3. 3

    Abecasis GR, Burt RA, Hall D, Bochum S, Doheny KF, Lundy SL et al. Genomewide scan in families with schizophrenia from the founder population of Afrikaners reveals evidence for linkage and uniparental disomy on chromosome 1. Am J Hum Genet 2004; 74: 403–417.

  4. 4

    JSSLG. Initial genome-wide scan for linkage with schizophrenia in the Japanese Schizophrenia Sib-Pair Linkage Group (JSSLG) families. Am J Med Genet 2003; 120B: 22–28.

  5. 5

    Bailer U, Leisch F, Meszaros K, Lenzinger E, Willinger U, Strobl R et al. Genome scan for susceptibility loci for schizophrenia. Neuropsychobiology 2000; 42: 175–182.

  6. 6

    Barr CL, Kennedy JL, Pakstis AJ, Wetterberg L, Sjogren B, Bierut L et al. Progress in a genome scan for linkage in schizophrenia in a large Swedish kindred. Am J Med Genet 1994; 54: 51–58.

  7. 7

    Blouin JL, Dombroski BA, Nath SK, Lasseter VK, Wolyniec PS, Nestadt G et al. Schizophrenia susceptibility loci on chromosomes 13q32 and 8p21. Nat Genet 1998; 20: 70–73.

  8. 8

    Brzustowicz LM, Hodgkinson KA, Chow EW, Honer WG, Bassett AS . Location of a major susceptibility locus for familial schizophrenia on chromosome 1q21–q22. Science 2000; 288: 678–682.

  9. 9

    Camp NJ, Neuhausen SL, Tiobech J, Polloi A, Coon H, Myles-Worsley M . Genomewide multipoint linkage analysis of seven extended Palauan pedigrees with schizophrenia, by a Markov-chain Monte Carlo method. Am J Hum Genet 2001; 69: 1278–1289.

  10. 10

    Coon H, Jensen S, Holik J, Hoff M, Myles-Worsley M, Reimherr F et al. Genomic scan for genes predisposing to schizophrenia. Am J Med Genet 1994; 54: 59–71.

  11. 11

    Coon H, Myles-Worsley M, Tiobech J, Hoff M, Rosenthal J, Bennett P et al. Evidence for a chromosome 2p13–14 schizophrenia susceptibility locus in families from Palau, Micronesia. Mol Psychiatry 1998; 3: 521–527.

  12. 12

    DeLisi LE, Mesen A, Rodriguez C, Bertheau A, LaPrade B, Llach M et al. Genome-wide scan for linkage to schizophrenia in a Spanish-origin cohort from Costa Rica. Am J Med Genet 2002; 114: 497–508.

  13. 13

    DeLisi LE, Shaw SH, Crow TJ, Shields G, Smith AB, Larach VW et al. A genome-wide scan for linkage to chromosomal regions in 382 sibling pairs with schizophrenia or schizoaffective disorder. Am J Psychiatry 2002; 159: 803–812.

  14. 14

    Devlin B, Bacanu SA, Roeder K, Reimherr F, Wender P, Galke B et al. Genome-wide multipoint linkage analyses of multiplex schizophrenia pedigrees from the oceanic nation of Palau. Mol Psychiatry 2002; 7: 689–694.

  15. 15

    Ekelund J, Lichtermann D, Hovatta I, Ellonen P, Suvisaari J, Terwilliger JD et al. Genome-wide scan for schizophrenia in the Finnish population: evidence for a locus on chromosome 7q22. Hum Mol Genet 2000; 9: 1049–1057.

  16. 16

    Fallin MD, Lasseter VK, Wolyniec PS, McGrath JA, Nestadt G, Valle D et al. Genomewide linkage scan for schizophrenia susceptibility loci among Ashkenazi Jewish families shows evidence of linkage on chromosome 10q22. Am J Hum Genet 2003; 73: 601–611.

  17. 17

    Faraone SV, Matise T, Svrakic D, Pepple J, Malaspina D, Suarez B et al. Genome scan of European-American schizophrenia pedigrees: results of the NIMH Genetics Initiative and Millennium Consortium. Am J Med Genet 1998; 81: 290–295.

  18. 18

    Garver DL, Holcomb J, Mapua FM, Wilson R, Barnes B . Schizophrenia spectrum disorders: an autosomal-wide scan in multiplex pedigrees. Schizophr Res 2001; 52: 145–160.

  19. 19

    Gurling H, Kalsi G, Brynjolfson J, Sigmundsson T, Sherrington R, Mankoo B et al. Genomewide genetic linkage analysis confirms the presence of susceptibility loci for schizophrenia, on chromosomes 1q32.2, 5q33.2, and 8p21–22 and provides support for linkage to schizophrenia, on chromosomes 11q23.3–24 and 20q12.1–11.23. Am J Hum Genet 2001; 68: 661–673.

  20. 20

    Hovatta I, Varilo T, Suvisaari J, Terwilliger JD, Ollikainen V, Arajarvi R et al. A genomewide screen for schizophrenia genes in an isolated Finnish subpopulation, suggesting multiple susceptibility loci. Am J Hum Genet 1999; 65: 1114–1124.

  21. 21

    Kaufmann CA, Suarez B, Malaspina D, Pepple J, Svrakic D, Markel PD et al. NIMH Genetics Initiative Millenium Schizophrenia Consortium: linkage analysis of African-American pedigrees. Am J Med Genet 1998; 81: 282–289.

  22. 22

    Lerer B, Segman RH, Hamdan A, Kanyas K, Karni O, Kohn Y et al. Genome scan of Arab Israeli families maps a schizophrenia susceptibility gene to chromosome 6q23 and supports a locus at chromosome 10q24. Mol Psychiatry 2003; 8: 488–498.

  23. 23

    Levinson DF, Mahtani MM, Nancarrow DJ, Brown DM, Kruglyak L, Kirby A et al. Genome scan of schizophrenia. Am J Psychiatry 1998; 155: 741–750.

  24. 24

    Lindholm E, Ekholm B, Shaw S, Jalonen P, Johansson G, Pettersson U et al. A schizophrenia-susceptibility locus at 6q25, in one of the world's largest reported pedigrees. Am J Hum Genet 2001; 69: 96–105.

  25. 25

    Macgregor S, Visscher PM, Knott SA, Thomson P, Porteous DJ, Millar JK et al. A genome scan and follow-up study identify a bipolar disorder susceptibility locus on chromosome 1q42. Mol Psychiatry 2004; 9: 1083–1090.

  26. 26

    Maziade M, Roy MA, Chagnon YC, Cliche D, Fournier JP, Montgrain N et al. Shared and specific susceptibility loci for schizophrenia and bipolar disorder: a dense genome scan in Eastern Quebec families. Mol Psychiatry 2004; 10: 486–499.

  27. 27

    Moises HW, Yang L, Kristbjarnarson H, Wiese C, Byerley W, Macciardi F et al. An international two-stage genome-wide search for schizophrenia susceptibility genes. Nat Genet 1995; 11: 321–324.

  28. 28

    Paunio T, Ekelund J, Varilo T, Parker A, Hovatta I, Turunen JA et al. Genome-wide scan in a nationwide study sample of schizophrenia families in Finland reveals susceptibility loci on chromosomes 2q and 5q. Hum Mol Genet 2001; 10: 3037–3048.

  29. 29

    Rees MI, Fenton I, Williams NM, Holmans P, Norton N, Cardno A et al. Autosome search for schizophrenia susceptibility genes in multiply affected families. Mol Psychiatry 1999; 4: 353–359.

  30. 30

    Schwab SG, Hallmayer J, Albus M, Lerer B, Eckstein GN, Borrmann M et al. A genome-wide autosomal screen for schizophrenia susceptibility loci in 71 families with affected siblings: support for loci on chromosome 10p and 6 [In Process Citation]. Mol Psychiatry 2000; 5: 638–649.

  31. 31

    Shaw SH, Kelly M, Smith AB, Shields G, Hopkins PJ, Loftus J et al. A genome-wide search for schizophrenia susceptibility genes. Am J Med Genet 1998; 81: 364–376.

  32. 32

    Sklar P, Pato MT, Kirby A, Petryshen TL, Medeiros H, Carvalho C et al. Genome-wide scan in Portuguese Island families identifies 5q31–5q35 as a susceptibility locus for schizophrenia and psychosis. Mol Psychiatry 2004; 9: 213–218.

  33. 33

    Stefansson H, Sigurdsson E, Steinthorsdottir V, Bjornsdottir S, Sigmundsson T, Ghosh S et al. Neuregulin 1 and susceptibility to schizophrenia. Am J Hum Genet 2002; 71: 877–892.

  34. 34

    Straub RE, MacLean CJ, Ma Y, Webb BT, Myakishev MV, Harris-Kerr C et al. Genome-wide scans of three independent sets of 90 Irish multiplex schizophrenia families and follow-up of selected regions in all families provides evidence for multiple susceptibility genes. Mol Psychiatry 2002; 7: 542–559.

  35. 35

    Wijsman EM, Rosenthal EA, Hall D, Blundell ML, Sobin C, Heath SC et al. Genome-wide scan in a large complex pedigree with predominantly male schizophrenics from the island of Kosrae: evidence for linkage to chromosome 2q. Mol Psychiatry 2003; 8: 695–705, 643.

  36. 36

    Williams NM, Rees MI, Holmans P, Norton N, Cardno AG, Jones LA et al. A two-stage genome scan for schizophrenia susceptibility genes in 196 affected sibling pairs. Hum Mol Genet 1999; 8: 1729–1739.

  37. 37

    Williams NM, Norton N, Williams H, Ekholm B, Hamshere ML, Lindblom Y et al. A systematic genomewide linkage study in 353 sib pairs with schizophrenia. Am J Hum Genet 2003; 73: 1355–1367.

  38. 38

    Lewis CM, Levinson DF, Wise LH, DeLisi LE, Straub RE, Hovatta I et al. Genome scan meta-analysis of schizophrenia and bipolar disorder, part II: schizophrenia. Am J Hum Genet 2003; 73: 34–48.

  39. 39

    Steiger JL, Russek SJ . GABAA receptors: building the bridge between subunit mRNAs, their promoters, and cognate transcription factors. Pharmacol Therapeutics 2004; 101: 259–281.

  40. 40

    Varecka L, Wu CH, Rotter A, Frostholm A . GABAA/benzodiazepine receptor alpha 6 subunit mRNA in granule cells of the cerebellar cortex and cochlear nuclei: expression in developing and mutant mice. J Comp Neurol 1997; 339: 341–352.

  41. 41

    Neelands TR, Macdonald RL . Incorporation of the pi subunit into functional gamma-aminobutyric Acid(A) receptors. Mol Pharmacol 1999; 56: 598–610.

  42. 42

    Hedblom E, Kirkness EF . A novel class of GABAA receptor subunit in tissues of the reproductive system. J Biol Chem 1997; 272: 15346–15350.

  43. 43

    Squires RF, Lajtha A, Saederup E, Palkovits M . Reduced [3H]flunitrazepam binding in cingulate cortex and hippocampus of postmortem schizophrenic brains: is selective loss of glutamatergic neurons associated with major psychoses? Neurochem Res 1993; 18: 219–223.

  44. 44

    Akbarian S, Huntsman MM, Kim JJ, Tafazzoli A, Potkin SG, Bunney Jr WE et al. GABAA receptor subunit gene expression in human prefrontal cortex: comparison of schizophrenics and controls. Cereb Cortex 1995; 5: 550–560.

  45. 45

    Huntsman MM, Tran BV, Potkin SG, Bunney Jr WE, Jones EG . Altered ratios of alternatively spliced long and short gamma2 subunit mRNAs of the gamma-amino butyrate type A receptor in prefrontal cortex of schizophrenics. Proc Natl Acad Sci USA 1998; 95: 15066–15071.

  46. 46

    Ohnuma T, Augood SJ, Arai H, McKenna PJ, Emson PC . Measurement of GABAergic parameters in the prefrontal cortex in schizophrenia: focus on GABA content, GABA(A) receptor alpha-1 subunit messenger RNA and human GABA transporter-1 (HGAT-1) messenger RNA expression. Neuroscience 1999; 93: 441–448.

  47. 47

    Lewis DA, Volk DW, Hashimoto T . Selective alterations in prefrontal cortical GABA neurotransmission in schizophrenia: a novel target for the treatment of working memory dysfunction. Psychopharmacology (Berlin) 2004; 174: 143–150.

  48. 48

    Wassef A, Baker J, Kochan LD . GABA and schizophrenia: a review of basic science and clinical studies. J Clin Psychopharmacol 2003; 23: 601–640.

  49. 49

    Coyle JT . The GABA-glutamate connection in schizophrenia: which is the proximate cause? Biochem Pharmacol 2004; 68: 1507–1514.

  50. 50

    Hashimoto T, Volk DW, Eggan SM, Mirnics K, Pierri JN, Sun Z et al. Gene expression deficits in a subclass of GABA neurons in the prefrontal cortex of subjects with schizophrenia. J Neurosci 2003; 23: 6315–6326.

  51. 51

    Ishikawa M, Mizukami K, Iwakiri M, Hidaka S, Asada T . GABAA receptor gamma subunits in the prefrontal cortex of patients with schizophrenia and bipolar disorder. Neuroreport 2004; 15: 1809–1812.

  52. 52

    Stefansson H, Sarginson J, Kong A, Yates P, Steinthorsdottir V, Gudfinnsson E et al. Association of neuregulin 1 with schizophrenia confirmed in a Scottish population. Am J Hum Genet 2003; 72: 83–87.

  53. 53

    Williams NM, Preece A, Spurlock G, Norton N, Williams HJ, Zammit S et al. Support for genetic variation in neuregulin 1 and susceptibility to schizophrenia. Mol Psychiatry 2003; 8: 485–487.

  54. 54

    Corvin AP, Morris DW, McGhee K, Schwaiger S, Scully P, Quinn J et al. Confirmation and refinement of an ‘at-risk’ haplotype for schizophrenia suggests the EST cluster, Hs.97362, as a potential susceptibility gene at the Neuregulin-1 locus. Mol Psychiatry 2004; 9: 208–213.

  55. 55

    Yang JZ, Si TM, Ruan Y, Ling YS, Han YH, Wang XL et al. Association study of neuregulin 1 gene with schizophrenia. Mol Psychiatry 2003; 8: 706–709.

  56. 56

    Tang JX, Chen WY, He G, Zhou J, Gu NF, Feng GY et al. Polymorphisms within 5' end of the Neuregulin 1 gene are genetically associated with schizophrenia in the Chinese population. Mol Psychiatry 2004; 9: 11–12.

  57. 57

    Li T, Stefansson H, Gudfinnsson E, Cai G, Liu X, Murray RM et al. Identification of a novel neuregulin 1 at-risk haplotype in Han schizophrenia Chinese patients, but no association with the Icelandic/Scottish risk haplotype. Mol Psychiatry 2004; 9: 698–704.

  58. 58

    Zhao X, Shi Y, Tang J, Tang R, Yu L, Gu N et al. A case control and family based association study of the neuregulin1 gene and schizophrenia. J Med Genet 2004; 41: 31–34.

  59. 59

    Petryshen TL, Middleton FA, Kirby A, Aldinger KA, Purcell S, Tahl AR et al. Support for involvement of neuregulin 1 in schizophrenia pathophysiology. Mol Psychiatry 2005; 10: 366–374.

  60. 60

    Straub RE, Jiang Y, MacLean CJ, Ma Y, Webb BT, Myakishev MV et al. Genetic variation in the 6p22.3 gene DTNBP1, the human ortholog of the mouse dysbindin gene, is associated with schizophrenia. Am J Hum Genet 2002; 71: 337–348.

  61. 61

    Schwab SG, Knapp M, Mondabon S, Hallmayer J, Borrmann-Hassenbach M, Albus M et al. Support for association of schizophrenia with genetic variation in the 6p22.3 gene, dysbindin, in sib-pair families with linkage and in an additional sample of triad families. Am J Hum Genet 2003; 72: 185–190.

  62. 62

    van den Oord EJ, Sullivan PF, Jiang Y, Walsh D, O'Neill FA, Kendler KS et al. Identification of a high-risk haplotype for the dystrobrevin binding protein 1 (DTNBP1) gene in the Irish study of high-density schizophrenia families. Mol Psychiatry 2003; 8: 499–510.

  63. 63

    Van Den Bogaert A, Schumacher J, Schulze TG, Otte AC, Ohlraun S, Kovalenko S et al. The DTNBP1 (dysbindin) gene contributes to schizophrenia, depending on family history of the disease. Am J Hum Genet 2003; 73: 1438–1443.

  64. 64

    Tang JX, Zhou J, Fan JB, Li XW, Shi YY, Gu NF et al. Family-based association study of DTNBP1 in 6p22.3 and schizophrenia. Mol Psychiatry 2003; 8: 1008.

  65. 65

    Williams NM, Preece A, Morris DW, Spurlock G, Bray NJ, Stephens M et al. Identification in 2 independent samples of a novel schizophrenia risk haplotype of the dystrobrevin binding protein gene (DTNBP1). Arch Gen Psychiatry 2004; 61: 336–344.

  66. 66

    Okada M, Corfas G . Neuregulin1 downregulates postsynaptic GABAA receptors at the hippocampal inhibitory synapse. Hippocampus 2004; 14: 337–344.

  67. 67

    Benson MA, Newey SE, Martin-Rendon E, Hawkes R, Blake DJ . Dysbindin, a novel coiled-coil-containing protein that interacts with the dystrobrevins in muscle and brain. J Biol Chem 2001; 276: 24232–24241.

  68. 68

    Peters MF, O'Brien KF, Sadoulet-Puccio HM, Kunkel LM, Adams ME, Froehner SC . Beta-dystrobrevin, a new member of the dystrophin family. Identification, cloning, and protein associations. J Biol Chem 1997; 272: 31561–31569.

  69. 69

    Knuesel I, Mastrocola M, Zuellig RA, Bornhauser B, Schaub MC, Fritschy JM . Short communication: altered synaptic clustering of GABAA receptors in mice lacking dystrophin (mdx mice). Eur J Neurosci 1999; 11: 4457–4462.

  70. 70

    Lo WS, Lau CF, Xuan Z, Chan CF, Feng GY, He L et al. Association of SNPs and haplotypes in GABAA receptor beta2 gene with schizophrenia. Mol Psychiatry 2004; 9: 603–608.

  71. 71

    Nurnberger Jr JI, Blehar MC, Kaufmann CA, York-Cooler C, Simpson SG, Harkavy-Friedman J et al. Diagnostic interview for genetic studies. Rationale, unique features, and training. NIMH Genetics Initiative. Arch Gen Psychiatry 1994; 51: 849–859; discussion 863–864.

  72. 72

    McGuffin P, Farmer A, Harvey I . A polydiagnostic application of operational criteria in studies of psychotic illness. Development and reliability of the OPCRIT system. Arch Gen Psychiatry 1991; 48: 764–770.

  73. 73

    Purcell S, Cherny SS, Sham PC . Genetic power calculator: design of linkage and association genetic mapping studies of complex traits. Bioinformatics 2003; 19: 149–150.

  74. 74

    Pritchard JK, Rosenberg NA . Use of unlinked genetic markers to detect population stratification in association studies. Am J Hum Genet 1999; 65: 220–228.

  75. 75

    Freedman ML, Reich D, Penney KL, McDonald GJ, Mignault AA, Patterson N et al. Assessing the impact of population stratification on genetic association studies. Nat Genet 2004; 36: 388–393.

  76. 76

    Sklar P, Schwab SG, Williams NM, Daly M, Schaffner S, Maier W et al. Association analysis of NOTCH4 loci in schizophrenia using family and population-based controls. Nat Genet 2001; 28: 126–128.

  77. 77

    Spitzer RL, Endicott J, Robins E . Research diagnostic criteria: rationale and reliability. Arch Gen Psychiatry 1978; 35: 773–782.

  78. 78

    Sklar P, Gabriel SB, McInnis MG, Bennett P, Lim YM, Tsan G et al. Family-based association study of 76 candidate genes in bipolar disorder: BDNF is a potential risk locus. Brain-derived neutrophic factor. Mol Psychiatry 2002; 7: 579–593.

  79. 79

    Barrett JC, Fry B, Maller J, Daly MJ . Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 2005; 21: 263–265.

  80. 80

    Lewontin RC . The interaction of selection and linkage. I. General considerations; heterotic models. Genetics 1964; 49: 49–67.

  81. 81

    Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B et al. The structure of haplotype blocks in the human genome. Science 2002; 296: 2225–2229.

  82. 82

    Hedrick PW, Thomson G . Maternal–fetal interactions and the maintenance of HLA polymorphism. Genetics 1988; 119: 205–212.

  83. 83

    Excoffier L, Slatkin M . Maximum-likelihood estimation of molecular haplotype frequencies in a diploid population. Mol Biol Evol 1995; 12: 921–927.

  84. 84

    Spielman RS, McGinnis RE, Ewens WJ . Transmission test for linkage disequilibrium: the insulin gene region and insulin-dependent diabetes mellitus (IDDM). Am J Hum Genet 1993; 52: 506–516.

  85. 85

    Rioux JD, Karinen H, Kocher K, McMahon SG, Karkkainen P, Janatuinen E et al. Genomewide search and association studies in a Finnish celiac disease population: Identification of a novel locus and replication of the HLA and CTLA4 loci. Am J Med Genet A 2004; 130: 345–350.

  86. 86

    Morris JA, Gardner MJ . Calculating confidence intervals for relative risks (odds ratios) and standardised ratios and rates. Br Med J (Clin Res Ed) 1988; 296: 1313–1316.

  87. 87

    Middleton FA, Pato CN, Gentile KL, McGann L, Brown AM, Trauzzi M et al. Gene expression analysis of peripheral blood leukocytes from discordant sib-pairs with schizophrenia and bipolar disorder reveals points of convergence between genetic and functional genomic approaches. Am J Med Genet B Neuropsychiatr Genet 2005; 136: 12–25.

  88. 88

    Falconer DS, MacKay TFC . Introduction to Quantitative Genetics. Prentice-Hall: Harlow, 1996.

  89. 89

    Mootha VK, Lindgren CM, Eriksson KF, Subramanian A, Sihag S, Lehar J et al. PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat Genet 2003; 34: 267–273.

  90. 90

    Reiner A, Yekutieli D, Benjamini Y . Identifying differentially expressed genes using false discovery rate controlling procedures. Bioinformatics 2003; 19: 368–375.

  91. 91

    Su AI, Cooke MP, Ching KA, Hakak Y, Walker JR, Wiltshire T et al. Large-scale analysis of the human and mouse transcriptomes. Proc Natl Acad Sci USA 2002; 99: 4465–4470.

  92. 92

    Su AI, Wiltshire T, Batalov S, Lapp H, Ching KA, Block D et al. A gene atlas of the mouse and human protein-encoding transcriptomes. Proc Natl Acad Sci USA 2004; 101: 6062–6067.

  93. 93

    DeRisi JL, Iyer VR, Brown PO . Exploring the metabolic and genetic control of gene expression on a genomic scale. Science 1997; 278: 680–686.

  94. 94

    Spellman PT, Sherlock G, Zhang MQ, Iyer VR, Anders K, Eisen MB et al. Comprehensive identification of cell cycle-regulated genes of the yeast Saccharomyces cerevisiae by microarray hybridization. Mol Biol Cell 1998; 9: 3273–3297.

  95. 95

    Hakak Y, Walker JR, Li C, Wong WH, Davis KL, Buxbaum JD et al. Genome-wide expression analysis reveals dysregulation of myelination-related genes in chronic schizophrenia. Proc Natl Acad Sci USA 2001; 98: 4746–4751.

  96. 96

    Mirnics K, Middleton FA, Marquez A, Lewis DA, Levitt P . Molecular characterization of schizophrenia viewed by microarray analysis of gene expression in prefrontal cortex. Neuron 2000; 28: 53–67.

  97. 97

    Vawter MP, Crook JM, Hyde TM, Kleinman JE, Weinberger DR, Becker KG et al. Microarray analysis of gene expression in the prefrontal cortex in schizophrenia: a preliminary study. Schizophr Res 2002; 58: 11–20.

  98. 98

    Torrey EF, Webster M, Knable M, Johnston N, Yolken RH . The Stanley Foundation brain collection and Neuropathology Consortium. Schizophr Res 2000; 44: 151–155.

  99. 99

    Turunen JA, Paunio T, Ekelund J, Suhonen J, Varilo T, Patrtonen T et al. Association of GABRG2 gene variants with susceptibility to schizophrenia. Am J Med Genet 2003; 122B: 18.

  100. 100

    Laurie DJ, Wisden W, Seeburg PH . The distribution of thirteen GABAA receptor subunit mRNAs in the rat brain. III. Embryonic and postnatal development. J Neurosci 1992; 12: 4151–4172.

  101. 101

    Wisden W, Laurie DJ, Monyer H, Seeburg PH . The distribution of 13 GABAA receptor subunit mRNAs in the rat brain. I. Telencephalon, diencephalon, mesencephalon. J Neurosci 1992; 12: 1040–1062.

  102. 102

    Mirnics K, Middleton FA, Lewis DA, Levitt P . Analysis of complex brain disorders with gene expression microarrays: schizophrenia as a disease of the synapse. Trends Neurosci 2001; 24: 479–486.

  103. 103

    Middleton FA, Mirnics K, Pierri JN, Lewis DA, Levitt P . Gene expression profiling reveals alterations of specific metabolic pathways in schizophrenia. J Neurosci 2002; 22: 2718–2729.

  104. 104

    Vawter MP, Ferran E, Galke B, Cooper K, Bunney WE, Byerley W . Microarray screening of lymphocyte gene expression differences in a multiplex schizophrenia pedigree. Schizophr Res 2004; 67: 41–52.

  105. 105

    Perl O, Ilani T, Strous RD, Lapidus R, Fuchs S . The alpha7 nicotinic acetylcholine receptor in schizophrenia: decreased mRNA levels in peripheral blood lymphocytes. FASEB J 2003; 17: 1948–1950.

  106. 106

    Tian J, Chau C, Hales TG, Kaufman DL . GABAA receptors mediate inhibition of T cell responses. J Neuroimmunol 1999; 96: 21–28.

  107. 107

    Erdo SL, Wolff JR . Gamma-aminobutyric acid outside the mammalian brain. J Neurochem 1990; 54: 363–372.

  108. 108

    Flames N, Long JE, Garratt AN, Fischer TM, Gassmann M, Birchmeier C et al. Short- and long-range attraction of cortical GABAergic interneurons by neuregulin-1. Neuron 2004; 44: 251–261.

Download references


We gratefully acknowledge the families and individuals who participated in this study. We thank Ana Dourado, Jose Valente, Carlos Paz Ferreira, Isabel Coelho, and MJ Soares for patient collection, Sinead O'Leary for analytical assistance, and Eric Lander, Ed Skolnick, Nick Patterson, and Vamsi Mootha for helpful discussions of the manuscript. We acknowledge the Stanley Medical Research Foundation, and Michael Knable, DO, Serge Weis, MD, Fuller Torrey, MD, Maree Webster, PhD, and Robert Yolken, MD for donating postmortem brain tissue. This research was supported by NIMH Grants MH52618 and MH058693 to CNP, and NARSAD Young Investigator awards to TLP and CNP. MJD was a Pfizer Fellow in Computational Biology.

Author information

Correspondence to P Sklar.

Additional information

Supplementary Information accompanies the paper on the Molecular Psychiatry website (

Supplementary information

Supplementary Table 1 (XLS 42 kb)

Supplementary Table 2 (XLS 98 kb)

Supplementary Table 3 (XLS 55 kb)

Supplementary Table 4 (XLS 42 kb)

Supplementary Table 5 (DOC 81 kb)

Rights and permissions

Reprints and Permissions

About this article


  • linkage disequilibrium
  • haplotypes
  • single nucleotide polymorphisms
  • oligonucleotide array sequence analysis
  • gene expression

Further reading