Analysis of the DISC1 translocation partner (11q14.3) in genetic risk of schizophrenia


Corresponding author: T. R. Merriman, Department of Biochemistry, University of Otago, Dunedin, New Zealand. E-mail:


The Disrupted-in-Schizophrenia 1 (DISC1) locus on human chromosome 1 was identified as a consequence of its involvement in a balanced translocation (1;11)(q42.1;q14.3) segregating with major psychiatric disorders in a Scottish family. Recently a comprehensive meta-analysis of genome-wide association scan data found no evidence that common variants of DISC1 (1q42.1) are associated with schizophrenia. Our aim was to test for association of variants in the 11q14.3 translocation region with schizophrenia. The 11q14.3 region was examined by meta-analysis of genome-wide scan data made available by the Genetic Association Information Network (GAIN) and other investigators (non-GAIN) through dbGap. P-values were adjusted for multiple testing using the false discovery rate (FDR) approach. There were no single-nucleotide polymorphisms (SNPs) significant (P < 0.05) after correction for multiple testing in the combined schizophrenia dataset. However, one SNP (rs2509382) was significantly associated in the male-only analysis with PFDR = 0.024. Whilst the relevance of the (1;11)(q42.1;q14.3) translocation to psychiatric disorders is currently specific to the Scottish family, genetic material in the chromosome 11 region may contain risk variants for psychiatric disorders in the wider population. The association found in this region does warrant follow-up analysis in further sample sets.

Schizophrenia has a significant heritable component of approximately 80% (Sullivan et al. 2003). Genome-wide association scans (GWASs) in schizophrenia have revealed the importance of rare structural variations in disease aetiology (International Schizophrenia Consortium 2008; Stefansson et al. 2008; Vacic et al. 2011; Walsh et al. 2008), however, these variants account for only a small proportion of cases (<5%). A genome-wide study testing common single-nucleotide polymorphism (SNP) variants for a role in the causation of schizophrenia provides evidence that the majority of the heritable component resides in thousands of risk variants of very small effect (International Schizophrenia Consortium et al. 2009).

The DISC1 locus on human chromosome 1 was identified as a consequence of its involvement in a balanced translocation (1;11)(q42.1;q14.3) segregating with major psychiatric disorders in a large Scottish family (Millar et al. 2000). The DISC1 gene encodes a scaffold protein involved in a number of pathways, such as neuronal migration and formation of the hippocampus, that have been implicated in the aetiology of schizophrenia (Hennah et al. 2006). Subsequent genetic association studies have implicated common variants of the DISC1 gene as risk factors for schizophrenia among Caucasian and Asian populations (Schumacher et al. 2009). However, a recent comprehensive meta-analysis of GWAS data, combined with imputation of data from 10 candidate gene studies, that consisted of 11 626 cases and 15 237 controls found no evidence that common variants of DISC1 are associated with schizophrenia (Mathieson et al. 2012).

Although a great deal of work has gone into assessing DISC1 for variants contributing to schizophrenia, genetic material in the translocation breakpoint (11q14.3, but originally described as 11q21) has been largely overlooked due to suggestions of an absence of genes in this region (Millar et al. 2000). However, the region of chromosome 11 that is involved in the translocation identified in the Scottish family (St Clair et al. 1990) is near the site of another reported reciprocal translocation between chromosome 9 and 11 that co-segregates with manic-depressive illness in an independent family (Smith et al. 1989). Recently, evidence of a transcript at the 11q14.3 breakpoint in the Scottish family with possible fusion proteins as a result of (1;11)(q42.1;q14.3) translocation has emerged (Zhou et al. 2008, 2010; Eykelenboom et al. 2012), leaving the chromosome 11 locus as an additional region of interest. Therefore, we examined the 11q14.3 region by meta-analysis of genome-wide scan data made available by the Genetic Association Information Network (GAIN) and other investigators (non-GAIN) through dbGap (GAIN Collaborative Research Group et al. 2007).

Subjects and methods


Data for sample sets of European-American (CEU) cases and matched controls were obtained from the general research use (GRU) subsets of two sample sets publically available through the dbGAP database (; GAIN schizophrenia [Genome-Wide Association Study of Schizophrenia (GAIN) dbGaP Study Accession: phs000021.v3.p2; cases = 1351 and controls = 1378] and non-GAIN schizophrenia [Molecular Genetics of Schizophrenia – non-GAIN Sample (MGS_non-GAIN) dbGaP Study Accession: phs000167.v1.p1; cases = 1149 and controls = 1347]. These sample sets are an initiative of the Foundation for National Institutes of Health (FNIH, and the US National Institute of Health (NIH) with support from various private partners (GAIN Collaborative Research Group et al. 2007). The data for both cohorts were originally obtained by a study conducted by Evanston North-Western Healthcare (Suarez et al. 2006) from the Molecular Genetics of Schizophrenia Collaboration dataset. Study participants were enrolled in these studies through recruitment in various medical institutions and support groups across USA and Australia. Diagnosis was performed by DIGS (Diagnostic Instrument for Genetic Studies) and DSM-IV criteria for schizophrenia or schizoaffective disorder. The study was approved by the appropriate ethics committee in addition to the relevant GAIN Data Access Committees and has therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. All persons gave their written informed consent prior to their inclusion in the study.

Single-nucleotide polymorphism selection and quality control

The two datasets were genotyped using the Affymetrix version 6.0 platform (GAIN Collaborative Research Group et al. 2007) and merged for combined analysis. The region surrounding the breakpoint on chromosome 11, the region q14.3 (position 88760352 to 92460352 according to NCBI37) containing 1038 SNPs that had been genotyped in both GAIN and non-GAIN datasets was analysed. Analysis only included SNPs within the defined region that had a minor allele frequency of >0.01 and a Hardy–Weinberg equilibrium P-value of >0.001, leaving 878 SNPs in the combined dataset. Single-nucleotide polymorphisms that contained more than two distinct alleles were excluded from the combined dataset in an effort to remove errors arising from coding alleles on different strands, leaving 874 SNPs. All SNPs in the table were checked for possible stranding issues by comparing the combined dataset allele frequencies to those in the two individual datasets. This process was also applied to the sex-specific datasets (see below) leaving 845 SNPs in the combined male dataset and 831 in the combined female dataset. A list of the genotyped SNPs used in this analysis can be found in Table S1. Single-nucleotide polymorphisms with experiment-wide significant association were visually inspected for genotyping integrity using the genotype cluster plots available through dbGAP or, in the case of the non-GAIN schizophrenia dataset, cluster plots were constructed using v4.0 Affymetrix Genotyping Console software. Individuals were filtered also, only those with a missing genotype proportion <0.2 were included in the analysis, resulting in a total 1123 cases (228 removed) and 1097 controls (281 removed) in the GAIN schizophrenia dataset, while all individuals in the non-GAIN dataset had missing genotype proportions of <0.2. The individuals from the GAIN and non-GAIN datasets were each split into male- and female-only data subsets. For the GAIN dataset, this gave a female subset of 329 cases and 570 controls and a male subset of 794 cases and 527 controls. Similarly for the non-GAIN data the female subset comprised 346 cases and 678 controls, while the male subset contained 803 cases and 669 controls.

Statistical analysis

Region-wide association analysis was undertaken using v1.07 plink (Purcell et al. 2007) in which both basic allelic association analysis and Cochran-Armitage Trend analysis were conducted. Imputation of the region was carried out using impute version 2 (Howie et al. 2009). The HapMap phase III plus the 1000 genomes data (June 2012, NCBI build 36) were used as reference. P-values from the allelic association analyses were adjusted for the multiple testing inherent in the analysis of the 874 SNPs in the 11q14.3 region using the false discovery rate (FDR) approach (Benjamini & Hochberg, 1995). Power calculations (Fig. S1) were performed as previously described (Johnson et al. 2001).


Analysis of the 11q14.3 region revealed the most associated SNP in the combined analysis to be rs7124944 (PUNCORRECTED = 2.3 × 10−4); however, it was not significant after correction for multiple testing by the FDR method (PFDR > 0.05) (Table 1). Data were split by sex due to previously reported sex-specific effects at DISC1 in various mental illnesses including schizophrenia (Chen et al. 2007; Hennah et al. 2003, 2008; Schumacher et al. 2009). One SNP (rs2509382) was significantly associated in the male-only analysis with PFDR = 0.024 (PUNCORRECTED = 4.8 × 10−5), with nominally significant association observed in each of the GAIN and non-GAIN datasets [Table S2; OR (95% CI) = 1.25 (1.01–1.56), P = 0.043 in GAIN and 1.46 (1.20–1.77), P = 1.7 × 10−4 in non-GAIN]. Significance was retained after correction for the sex analysis (P = 0.048). SNP rs2509382 was in strong LD (r2 = 0.90) with the other four nominally significant SNPs in the male-only analysis (Table 1). SNP rs2509382 maps 131 kb upstream of the identified gene [DISC1 fusion partner (DISC1FP)] in this region, with the remaining four SNPs mapping between rs2509382 and the beginning of DISC1FP (84–8 kb upstream). No SNPs in the female-only analysis were significantly associated after correction for multiple testing.

Table 1. Markers (top 5) with strongest evidence for association in combined GAIN and non-GAIN schizophrenia and separate male and female analyses
SNP 111222Minor alleleaOR [95% CI]PUNCORRECTEDPFDR
  1. a

    2 = minor allele; 1 = major allele.

  2. b

    Genotype integrity was verified using the genotype cluster plots available through dbGAP or, in the case of the non-GAIN schizophrenia dataset, cluster plots were constructed using Affymetrix Genotyping Console software.

Rs7124944Case1700 (0.748)531 (0.234)41 (0.018)613 (0.135)1.26 [1.11–1.43]2.3 × 10−40.097
 Control1936 (0.792)478 (0.196)30 (0.012)538 (0.110)
Rs1458326Case1127 (0.496)921 (0.406)222 (0.098)1365 (0.301)0.87 [0.79–0.95]1.2 × 10−30.15
 Control1100 (0.450)1067 (0.437)277 (0.113)1621 (0.332)
Rs1674086Case1080 (0.481)936 (0.417)227 (0.101)1390 (0.310)0.87 [0.79–0.95]1.3 × 10−30.15
 Control1033 (0.431)1091 (0.455)272 (0.114)1635 (0.341)
Rs12791202Case1469 (0.648)707 (0.312)90 (0.040)887 (0.196)1.18 [1.07–1.32]1.6 × 10−30.15
 Control1674 (0.687)695 (0.285)68 (0.028)831 (0.171)
Rs1792361Case1097 (0.483)940 (0.414)235 (0.103)1410 (0.310)0.87 [0.80–0.95]1.6 × 10−30.15
 Control1065 (0.436)1092 (0.447)287 (0.117)1666 (0.341)
Rs2509382 bCase1074 (0.673)466 (0.292)57 (0.036)580 (0.182)1.35 [1.17–1.57]4.8 × 10−50.024
 Control886 (0.741)283 (0.237)27 (0.023)337 (0.141)
Rs35003084Case1050 (0.662)485 (0.306)52 (0.033)589 (0.186)1.34 [1.16–1.55]8.2 × 10−50.024
 Control862 (0.729)296 (0.250)24 (0.020)344 (0.146)
Rs12787172Case1055 (0.661)483 (0.302)59 (0.037)601 (0.188)1.33 [1.15–1.53]1.1 × 10−40.024
 Control867 (0.725)302 (0.253)27 (0.023)356 (0.149)
Rs1404531Case1048 (0.657)490 (0.307)58 (0.036)606 (0.190)1.32 [1.14-1.52]1.6 x 10-40.027
 Control863 (0.723)301 (0.252)30 (0.025)361 (0.151)
Rs11019229Case1052 (0.666)467 (0.296)61 (0.039)589 (0.186)1.31 [1.13–1.51]2.9 × 10−40.040
 Control858 (0.726)295 (0.250)29 (0.025)353 (0.149)
Rs16916137Case298 (0.884)38 (0.113)1 (0.003)40 (0.059)1.95 [1.25–3.03]2.8 × 10−30.998
 Control628 (0.939)40 (0.060)1 (0.001)42 (0.031)
Rs56287704Case544 (0.806)125 (0.185)6 (0.009)137 (0.102)1.40 [1.11–1.77]4.0 × 10−30.998
 Control1069 (0.857)172 (0.138)7 (0.006)186 (0.075)
Rs1531249Case174 (0.258)347 (0.514)154 (0.228)655 (0.485)1.20 [1.05–1.37]7.1 × 10−30.998
 Control394 (0.316)610 (0.489)244 (0.196)1098 (0.440)
Rs7103426Case646 (0.957)29 (0.043)0 (0.000)29 (0.021)2.00 [1.18–3.41]8.4 × 10−30.998
 Control1222 (0.979)25 (0.020)1 (0.001)27 (0.011)
Rs10501718Case643 (0.957)29 (0.043)0 (0.000)29 (0.022)2.00 [1.18–3.40]8.7 × 10−30.998
 Control1213 (0.979)25 (0.020)1 (0.001)27 (0.011)

Population stratification was assessed using v1.0 eigenstrat (Li & Yu, 2008), with a genomic inflation factor of 1.15 in the combined males and females analysis and 1.11 in the male-only analysis. Given that these factors are close to that typically expected from GWAS (<1.1; Yang et al. 2011) and that the minor allele frequency of rs2509382 is very similar in diverse 1000 Genomes populations [European (EUR) = 0.161, African (AFR) = 0.152, Asian (ASN) = 0.124] it is unlikely that population stratification is playing a significant role in our data.

Given that the variants genotyped on the Affymetrix version 6.0 array capture, on average, 83% of the common variation in the Caucasian genome (Li et al. 2008) we imputed the combined male-only and combined female-only datasets. In the male-only dataset, an additional 19 SNPs were found to be nominally associated with schizophrenia that were in strong LD with rs2509382 (r2 > 0.86; 1.29 < OR < 1.35; 8.1 × 10−4 < P < 4.8 × 10−5). The 24 SNPs formed a clear haplotype block and all were mapped between 131 and 1 kb upstream of DISC1FP1. Of these SNPs, rs2509382 at the extreme downstream end of the block was the strongest associated (+131 kb); with the third strongest associated (rs34339130: OR = 1.43, P = 5.1 × 10−5, r2 with rs2509382 of 0.86) at the extreme upstream end of the block (+1 kb). No significant associations were observed in the imputed female-only dataset.


According to the genomic fusion sequence described in the original paper (St Clair et al. 1990) the region of 11q14.3 containing the proposed DISC1FP transcript spans 89984400–90648220 bp (NCBI37) with the break occurring between bases 90361098 and 90361099. There are two possible fusion transcripts created by DISC1FP and DISC1 in the Scottish family, the first of which is a fusion of exons 1–3 of the DISC1FP gene with exons 9–13 of DISC1 and the second is a fusion of DISC1 exons 1–8 with exons 4–7 of DISC1FP (Zhou et al. 2008). Given the structure of the fusion proteins, the nominally associated SNPs reported here are outside of the actual gene region. This does not, however, preclude a possible role for these SNPs in common schizophrenia – SNPs that influence gene expression and are causative for common disease have been reported in intergenic regions (Cunnington et al. 2010; Carvajal-Carmona et al. 2011). Of course, whilst the relevance of this fusion to psychiatric disorders is, at this stage, specific to the Scottish family, the fusion implicates genetic material in the chromosome 11 region as a candidate for harbouring risk variants for psychiatric disorders in the wider population. The association found in this region can be regarded as potentially genuine and does warrant follow-up analysis in further sample sets.

Concerning a possible role for DISC1FP1 in schizophrenia, Zhou and colleagues (2010) identify that within the human brain there is novel expression of a single exon of DISC1FP1 that is not present in lymphoblastoid cells; additionally, there are several splice variations of the gene present in human thalamus. There is some suggestion that this gene may have an inhibitory interaction with the brain expressed CHORDC1 which has been implicated in stress response. Zhou et al. (2010) found that the DISC1–DISC1FP1 fusion protein is insoluble, in contrast to the DISC1FP1–DISC1 fusion protein. Insoluble DISC1 proteins have been found to aggregate in the brains of patients with schizophrenia and it is suggested that this limits the ability of DISC1 to interact with necessary binding partners (Leliveld et al. 2008). The DISC1–DISC1FP1 proteins cluster in the mitochondria where they destroy membrane potential (Eykelenboom et al. 2012).

Despite the amount of data analysed in this study (>800 genotyped SNPs in a total of 2272 schizophrenia cases and 2444 controls), two questions remain that can only be addressed by analysis of larger sample sets with more genetic variants. First, in the case of common risk variants, we were underpowered to detect weak effects of OR < 1.3 in the combined and male-only analyses and power was very low in the female-only analysis (Fig. S1). However, we are confident that we have excluded the possibility of the existence of a schizophrenia variant with modest effect (OR > 1.3) in the analysed region. Second, our study could not address the possible contribution of rare variants in the region to schizophrenia. Although variants were tested to a level of MAF 0.01, our power was minimal at this frequency (Fig. S1).


R.D. was supported by a University of Otago Postgraduate Scholarship. We acknowledge the NIH GWAS data repository. Funding support for the Genome-Wide Association Study of Schizophrenia (‘GAIN’) and Molecular Genetics of Schizophrenia (MGS) – non-GAIN sample (‘non-GAIN’) was provided by Genomics Research Branch at NIMH (see below) and the genotyping and analysis of samples was provided through the Genetic Association Information Network and under the MGS U01s: MH79469 and MH79470. Assistance with data cleaning was provided by the National Center for Biotechnology Information. The MGS dataset(s) used for the analyses described in this manuscript were obtained from the database of Genotype and Phenotype (dbGaP) found at through dbGaP accession numbers phs000021.v3.p2 (GAIN) and phs000167.v1.p1 (non-GAIN). Samples and associated phenotype data for the MGS GWAS study were collected under the following grants: NIMH Schizophrenia Genetics Initiative U01s: MH46276 (CR Cloninger), MH46289 (C Kaufmann), and MH46318 (MT Tsuang); and MGS Part 1 (MGS1) and Part 2 (MGS2) R01s: MH67257 (NG Buccola), MH59588 (BJ Mowry), MH59571 (PV Gejman), MH59565 (Robert Freedman), MH59587 (F Amin), MH60870 (WF Byerley), MH59566 (DW Black), MH59586 (JM Silverman), MH61675 (DF Levinson), and MH60879 (CR Cloninger). None of the authors declare conflict of interest.