Introduction

Genetic components play an important role in the etiology of schizophrenia based on evidence of family, twin, and adoption studies, with an estimated heritability of 60–80% [1]. Previous genetic linkage studies have identified chromosome 1q42 as a candidate region for schizophrenia [2, 3]. A balanced translocation (1; 11) (q42.1; q14.3) disrupting two genes (disrupted-in-schizophrenia 1 [DISC1] and DISC2) in chromosome 1 was found to be associated with major mental illnesses, including schizophrenia in a large Scottish family pedigree [4]. Although earlier small-scale genetic association studies have reported modestly significant associations between DISC1 and schizophrenia [5], more recent large-scale meta-analyses [6, 7] and the largest genome-wide association study (GWAS) [8] have concluded that DISC1 is not valid as a common risk gene for the Diagnostic and Statistical Manual of Mental Disorders (DSM)-diagnosed schizophrenia, especially in Caucasian ancestry. A recent meta-analysis [9], containing a larger Chinese and Japanese population than in previous studies and including 31 eligible studies with 13 markers in DISC1 (the maximal case number was 9031 and control number 9733 for rs821616) suggested that DISC1 polymorphisms (rs821616 and rs821597) increased schizophrenia risk in overall populations and especially in Chinese populations.

Although DISC1 is not associated with DSM-diagnosed schizophrenia, accumulating evidence has been reported in association between DISC1 and specific endophenotypes that are commonly associated with schizophrenia and other major mental disorders, including some neurocognitive endophenotypes (such as sustained attention [10], working memory [11, 12]), electrophysiological endophenotypes (such as auditory-event-related potential [13]), and neuroimage endophenotypes (such as reduced gray matter volume of hippocampus [14], prefrontal cortex [12], cingulate and several other brain regions [15], and white matter integrity [16]).

DISC1 is an intracellular hub for protein/RNA networks, involving the function for early neurodevelopment and synaptic regulation [17], including progenitor proliferation [18], neuronal migration [18, 19], neurites outgrowth [20], synaptic formation and regulation [21, 22], and neuronal signaling [22, 23]. Many lines of evidence have indicated that deficits of these neuronal processes, in particular those in early development, underlie circuitry and behavioral endophenotypes relevant to a wide range of mental illness [24]. Several DISC1 interactors have also been defined as independent genetic susceptibility factors for psychiatric illness [25]. Various DISC1 transgenic animal models coupled with different environmental stressors, have proven successful in satisfying face validity as models of a wide range of human psychiatric conditions [26].

The alternative splicing of DISC1 in the human brain is exceptionally complex, with more than 50 splice variants in different stages of brain development [27]. However, results of studies of the mRNA level of total amount of different DISC1 isoforms were inconsistent between schizophrenia patients and controls. One study reported no significant differences in the dorsal lateral prefrontal cortex [28], another reported significantly increased total DISC1 in the mononuclear cells of peripheral blood in drug-naive schizophrenia patients [29], and the other study reported significantly decreased DISC1 in the whole blood of schizophrenia patients [30]. An extensive study of mRNA levels of different DISC1 isoforms found that the expressions of three isoforms, including two short isoform variants and an extra-short isoform variant (Esv1), were significantly higher in the hippocampus of patients with schizophrenia [27].

Although most GWAS and meta-analyses showed negative results for associations between DISC1 and schizophrenia, several factors motivated our group to study this candidate gene. First, as shown in the most recent meta-analysis including more Chinese and Japanese subject [9], ancestral differences may influence the association evidence of this gene. Second, DISC1 is reported to be associated with some neurocognitive and neuroimaging features of schizophrenia, and is significantly associated with specific subgroups of schizophrenia, which tests the hypothesis that this gene may be specifically associated with certain subgroups of schizophrenia or neurocognitive endophenotypes. Third, convergent functional genomics approaches that integrate genetic, gene expression, and functional data continue to support the involvement of DISC1 in schizophrenia and related biological pathways [31].

In our previous study [10], we found significant associations between DISC1 and a subgroup of schizophrenia patients with deficient sustained attention assessed by the continuous performance test (CPT), using 102 families with at least two siblings affected with schizophrenia. In the present study, we aimed to re-sequence the genomic regions of this gene and to study the associations between the identified single-nucleoride polymorphisms (SNPs) and the subgroup of schizophrenia patients with deficient sustained attention, and to study the mRNA expression of different isoforms of DISC1.

Methods

Subjects

Subjects for direct sequencing

Study subjects included the probands from a project, in which schizophrenia probands from both simplex (i.e., only one affected offspring in the two-generation family) and multiplex families (i.e., at least two affected siblings in the family) were recruited from National Taiwan University Hospital and Ju-Shan Psychiatric Hospital from 2002 to 2005. The study protocol was reviewed and approved by the institutional review boards of the two hospitals. The inclusion criteria were probands meeting the DSM, 4th edition (DSM-IV) diagnostic criteria of schizophrenia and the exclusion criteria were severe neurological abnormality, substance-use disorder except for smoking, intellectual disability, and aboriginal ancestry. Normal controls were recruited with inclusion criteria of no lifetime psychiatric disorders and no family history of schizophrenia in first-degree relatives. Signed informed consent was obtained from all subjects after detailed explanation of the study. A total of 50 independent multiplex patients, 50 simplex patients, and 50 normal controls were selected for direct sequencing study.

Subjects for genotyping

The patients came from two types of families: multiplex families and simplex families. The multiplex families were recruited from two research programs: the first one [32] conducted from 1993 to 2001 and the second one [33] conducted from 1998 to 2002. Protocols of the research projects were reviewed and approved by the institutional review board of National Taiwan University Hospital. Signed informed consent was obtained from family members after the procedures had been fully explained. The 93 families of the first research program were interviewed by research psychiatrists using the Psychiatrist Diagnostic Assessment [34]. The 609 families of the second research program were interviewed by well-trained assistants using the Mandarin Chinese version of the Diagnostic Interview for Genetic Studies [35]. For both studies, the final research diagnosis was formulated by integrating interview data with clinical information from medical records, based on DSM-IV criteria. To assure the independence of multiplex subjects, only one affected sibling was randomly selected from each multiplex family. A total of 702 patients from these multiplex families were included in this study. A total of 87 independent patients from the multiplex families were included in our previous family-based association [10]. The simplex families were recruited from another independent project [36]. The diagnostic assessment was the same as that for the second research project. A total of 210 simplex families were recruited. Therefore, a total of 912 schizophrenia patients comprised the analytic sample for this genotyping study.

The normal controls were selected from a representative Taiwanese supernormal genomic sample [37] with the inclusion criteria of age over 60 years and a score above 14 on the Short Portable Mental Status Questionnaire. Control subjects were recruited by the Institute of Biomedical Science, Academia Sinica. A total of 600 supernormal control individuals (313 males and 287 females) were included in this study.

Subjects for gene expression study

The subjects for the mRNA expression study were from the same project as the direct sequencing study. A total of 85 patients and 36 controls were included in this gene expression study. The mean age of controls was 36.9 (±10.7) years and 41.7% were male. The mean age of patients was 33.0 (±9.1) years and 58.8% were male. No significant differences were found in the age and gender of patients and controls.

Methods

Direct sequencing

All exons, promoter regions, and the associated haplotype regions as reported in our previous study [10] were defined using the following bioinformatics process. As the highly conserved regions across different species are more resistant to evolutionary pressure, implying more important for gene function, we also sequenced the highly conserved regions.

The exon regions of DISC1 were evaluated for all transcripts and variants of expressed sequence tags updated by the Integrated Splicing Variants database [38]. Six isoforms, including long form (L), long form variants (Lv), short form (S), extra-short form (Es), the other two extra-short isoforms, and six expressed sequence tags variants were found in the DISC1 genomic region (Supplementary Fig. 1). The promoter region was defined as the 2000 base pairs (bp) before the start exon of each transcript. The highly conserved regions were predicted by MultiPipMaker (http://pipmaker.bx.psu.edu/pipmaker/) output results. These conserved regions were arranged into clusters by MultiPipmaker through the percent identity plot among humans, chimpanzees, mice, and rats. The genomic positions of previously significantly associated haplotype regions from rs2793092 to rs2793091 (from intron 3 to intron 4 according to L isofrom) [10] and highly conserved regions were integrated.

The genomic regions of DISC1 selected from the above bioinformatics procedures were amplified by PCR. PCR products were purified with exonuclease I and shrimp alkaline phosphatase (USB Corporation, Cleveland, Ohio, USA), and sequenced from both ends. DNA-sequencing reactions were performed with BigDye Terminator Cycle Sequencing Version 3.1 (Applied Biosystems, Foster City, CA, USA) followed by analysis on an ABI 3730xl DNA Analyzer (Applied Biosystems, Foster City, CA, USA).

The resulting sequences were compared and aligned using the Polyphred Sequence Alignment Editor (http://droog.mbt.washington.edu/PolyPhred.htm). The reference sequences were obtained from the National Center for Biotechnology Information website (http://www.ncbi.nlm.nih.gov/).

SNP genotyping

The selection criteria for further large-scale genotyping of genetic variants identified from direct sequencing are (1) all exon variants; (2) differences in minor allele frequency (MAF) >2% between controls and patients, if the variant frequency is less common (MAF in these subjects was between 0 and 10%); and (3) MAF differences >4% for more common variants (MAF between 10 and 50%). Genotyping was performed using the matrix-assisted laser desorption/ionization–time of flight (MALDI-TOF) mass spectrometry. Primers and probes flanking the SNPs were designed using SpectroDESIGNER software (Sequenom, San Diego, CA, USA). A DNA fragment (100–300 bp) encompassing the SNP site was amplified using PCR (GeneAmp 9700 thermocycler, Applied Biosystems, USA) according to the manufacturer’s instructions.

After removing the un-incorporated deoxynucleotide triphosphate (dNTP) and inactivating the shrimp alkaline phosphatase from the PCR product, primer extension was performed by adding the probe, Thermo Sequenase (Amersham Pharmacia, Piscataway, NJ, USA), and an appropriate dideoxynucleotide triphosphate/dNTP mixture, followed by 55 cycles of denaturing at 94 °C for 5 s, annealing at 52 °C for 5 s, and extension at 72 °C for 5 s. The other extension products were differentiated by mass through MALDI-TOF.

Real-time PCR (RT-PCR)

The mRNA expressions of different DISC1 isoforms, including total, L, Lv, S, Es, and Esv1 (according to the nomenclature in Nakata et al. [27]), were measured in Epstein–Barr virus (EBV)-transformed lymphoblastic cell lines. Trizol Reagent (Invitrogen Life Technologies) was used according to the manufacturer’s guidelines to extract total RNA.

RT-PCR was performed for DISC1 isoforms and a housekeeping gene, TATA-box binding protein (TBP), using pre-designed gene-specific TaqMan® probes and primer sets purchased from Applied Biosystems (Branchburg, NJ). The sequences of primers of different isoforms were listed in Supplementary Table 1. RT-PCR amplification was conducted using Taqman One-Step RT-PCR Master Mix Reagent (Applied Biosystems, Branchburg, NJ) on an ABI StepOne Plus System (Applied Biosystems, Branchburg, NJ), according to the manufacturer’s instructions. Gene expression was quantified relative to TBP expression using StepOne Software (Applied Biosystems, Branchburg, NJ) and the relative quantification method. The relative expression level of DISC1 isoforms compared with that of TBP was defined as −ΔCT = −[CTDISC1 − CTTBP], where CT was the cycle threshold. The DISC1 mRNA/TBP mRNA ratio was calculated from 2−ΔCT × K, in which K was a constant.

Luciferase reporter assay

To discern the potential effects of different alleles of the 3′-untranlated region (UTR) variant (rs11122324) of DISC1 Es and Esv1 isoforms, two luciferase reporter vectors were constructed by cloning different alleles (G-allele and A-allele) of rs11122324 in the 3′-UTR region of DISC1 Es (also Esv1) isoform into pGL4.10[luc2] vector (Promega, Madison City, WI, USA). The cytomegalovirus promoter was ligated to drive the luciferase expression (luc2) and the original 3′-UTR SV40 was replaced with the 3′-UTR sequence of DISC1 Es (also Esv1) isoform. All inserts were confirmed by direct sequencing. The construction process is displayed in Supplementary Fig. 2. The 3′-UTR fragment with A-allele of rs11122324 was amplified by PCR from the genomic DNA of a patient confirmed to carry this allele. The 3′-UTR fragment with G-allele was obtained by single point mutagenesis (A to G) from the above-described fragment. Therefore, the two vectors were the same, except for the variant of rs11122324 in their 3′-UTR. We transfected three vectors, the first is control vector (pCMV4.10), the G-allele vector (pCMV4.10-DISC1-G), and the A-allele vector (pCMV4.10-DISC1-A) into two cell lines (HEK293T and U251), respectively. Cells were seeded in 24-well plates at a density of 1 × 105 cells per well and cultured in Dulbecco’s modified Eagle’s medium (Invitrogen, Carlsbad City, CA, USA) supplemented with 10% fetal bovine serum. Cells were co-transfected with 1.8 μg of each reporter vector and 0.2 μg of Renilla luciferase plasmid (Promega) using Lipofectamine 2000 transfection reagent (Invitrogen). The cells were collected after 72 h. Luciferase activity was immediately detected by the Dual-Luciferase Reporter Assay System (Promega) using a GloMax 96 Luminometer (Promega) and luciferase assay kit (Promega). The experiments were repeated at least three times in both cell lines. Relative luciferase activity is calculated as the ratio of Firefly luciferase activity to Renilla luciferase activity. For control of the variations of transfection efficacy in repeated experiments, the relative transcription activities of the A-allele and the G-allele vector were calculated as the ratio of relative luciferase activity of the A-allele or the G-allele vector to that of the control vector (pCMV4.10) in the same experiment.

Sustained attention assessment

A CPT machine from Sunrise System, v. 2.20 (Pembroke, MA, USA) was used to assess sustained attention. The procedure has satisfactory reliability and has been described in detail elsewhere. Each subject undertook two CPT sessions: the undegraded 1–9 task and the 25% degraded 1–9 task. Subjects were asked to respond whenever the number “9” preceded by the number “1” appeared on the screen. During the 25% degraded session, a pattern of “snow” was used to partially obscure the background and foreground, visually distorting the number. One signal-detection index of performance on the test, sensitivity (d′), was derived from the hit rate (probability of response to target trials) and false-alarm rate (probability of response to non-target trials). Sensitivity (d′) is an individual’s ability to discriminate target stimuli from non-target stimuli.

The z-score of d′ on CPT was calculated according to age, gender, and education based on a community-based sample of 345 controls [35]. The cut point of d′ was set as −2.5 from our previous family genetic study [39]. In this study, 303 patients had z-scores of d′ on undegraded CPT that were below −2.5 (deficit) and 458 patients with scores above −2.5 (non-deficit); 377 patients had z-scores of d′ on degraded CPT that were below −2.5 (deficit) and 360 patients with scores above −2.5 (non-deficit).

Statistical analysis

Hardy–Weinberg equilibrium was assessed by using the ALLELE procedure in SAS/GENETICS release 9.4. Allele and genotype frequencies were compared between the case and control groups by using the χ2-test or Fisher’s exact test where appropriate. The false discovery rate method was used to correct for multiple comparisons. The relative expression levels of different DISC1 isoforms between the case and control groups were compared using the two-sample t-test. The relative transcription activities (ratio to the control vector (pCMV4.10)) of the G-allele and A-allele vector were compared using paired t-test. A p-value < 0.05 was considered significant.

Results

Genetic association results

A total of 24,854 bp of DISC1 genomic region have been sequenced. Using direct sequencing, 83 genetic variations have been identified, of which 16 variants are not found in 1000 Genomes or other public variation resources, and 10 are exon variants, of which none is novel. All genetic variations are listed in Supplementary Table 2. Sixteen variations, including 10 exon variants, met the criteria for further genotyping in 600 controls and 912 schizophrenia patients. However, we failed to genotype two exon variants (P193S and R264Q) using our genotyping method. Therefore, a total of 14 SNPs had the genotyping data. All 14 SNPs were compatible with Hardy–Weinberg equilibrium. The detailed information and association results in all samples are listed in Table 1. No significant differences were found between all patients (N = 912) and controls in the allele and genotype frequencies for the 14 SNPs. As our previous association results were found to be significant in the multiplex families, patients were further subgrouped into the multiplex patients group (N = 702) and the simplex patient group (N = 210). The genotype frequency of rs11122324 was found at the 3′-UTR of Es and Esv1 isoform, and that of rs2793091 at intron 4 was significantly different between the multiplex patients and controls (corrected p = 0.0146 and 0.0036, respectively). The allele frequencies of the two SNPs were not significantly different after multiple correction. No significant differences were found between the multiplex patients and controls in the allele and genotype frequency for the other 12 SNPs. No significant differences in either allele or genotype frequencies were found between the simplex patients and controls.

Table 1 The detail information and association results of the 14 genotyped genetic variants identified from direct sequencing in all samples (N = 1512)
Full size table

Further subgrouping of the multiplex patients into those with and without sustained attention deficit revealed that the genotype frequencies of rs11122324 and rs2793091 were significantly different between the patients with deficient sustained attention and controls (corrected p = 0.0146 and 0.0036, respectively, in undegraded CPT and corrected p = 0.0058 and 0.0013, respectively, in degraded CPT), but were not significantly different between the multiplex patients without sustained attention deficit and controls. The allele frequencies of the two SNPs were not significantly different after multiple correction. The detailed genotype distribution and genotype-wise association results of rs11122324 and rs2793091 in the different subgroups are listed in Table 2. The proportions of AA genotype of rs11122324 and rs2793091 (recessive genotype) were observed to be increased in the subgroups of multiplex and deficient sustained attention. The genotype-wise significance became even more evident when compared under the recessive model (AA genotype vs. non-AA genotype) of the two SNPs. The detailed results are listed in Table 2.

Table 2 The genotype distribution and genotype-wise association results of rs11122324 and rs2793091 under additive and recessive model in the different subgroups
Full size table

We further explored the association between the DISC1 SNPs and sustained attention in schizophrenia patients. We found significant association between the performance of CPT and the genotypes of rs1865225, rs11122324, rs2793091, and rs2492367 (Supplementary Table 3). However, the association results were not significant after multiple correction. Under recessive models of rs11122324 and rs2793091, it showed the patients with AA genotypes of rs11122324 and rs2793091 had significant poorer performance of CPT than those with GG and GA genotypes (t = 2.430, d.f. = 757, p = 0.015 for rs11122324; t = 2.894, d.f. = 706, p = 0.004 for rs2793091).

Expression study

The mRNA expression levels of DISC1 Es and Esv1 isoform in the EBV-transformed lymphocytes of schizophrenia (N = 85) were found to be significantly higher than those of controls (N = 36) (t = 2.03, d.f. = 119, p = 0.047 for Es isoform; t = 2.13, d.f. = 119, p = 0.035 for Esv1 isoform), but the results became nonsignificant after multiple correction. The mRNA expression levels of total DISC1, L, Lv, S isoforms were not significantly different between patients and controls (Fig. 1a). As rs11122324 was located at the 3′-UTR of Es and Esv1 isoforms, the relationship between the genotypes of rs11122324 and the expression of the two isoforms were explored. Patients with A-allele of rs11122324 (GA and AA genotypes) were found to have significantly higher expression of Es isoform than those with GG genotype (dominant model for A-allele) (t = 2.01, d.f. = 76.9, p = 0.048). There were no significant differences for Esv1 isoform expression under this model. However, there were no significant differences for both Es and Esv1 isoform between patients with AA genotype of rs11122324 and those with GA and GG genotypes (recessive model) (Fig. 1b).

Fig. 1
figure 1

Gene expression study of different DISC1 isoforms in the EBV-transformed lymphoblastic cells of schizophrenia patients and controls. a Relative transcriptional levels of the different DISC1 isoforms in the EBV-transformed lymphoblastic cells of schizophrenia patients (n = 85) and controls (n = 36). Data are normalized by the means of the control group and represented as mean ± SE. An asterisk denotes a significant difference (p < 0.05). b The comparisons between the genotypes of rs11122324 and the relative transcriptional levels of Es and Esv1 isoform. Data are represented as mean ± SE. An asterisk denotes a significant difference (p < 0.05). The level of Es isoform in patients with the AA (n = 12) and GA (n = 43) genotype is significantly higher than that in patients with the GG genotype (n = 24) using independent t-test. The level of Es isoform in patients with the AA genotype (n = 12) is not significantly different from that in patients with the GG (n = 43) and GA genotype (n = 24) using Mann–Whitney U-test. The comparisons of the levels of Esv1 isoform between patients with different genotypes are not significant

Full size image

We found there were no significant differences of all the isoforms among the simplex patients, the multiplex patients and control group. There were significantly negative correlations between the performances of CPT and the expression of Esv1 isoform in all samples (Pearson correlation r = −0.255, p = 0.005, d.f. = 121, with the z-scores of undegraded CPT; r = −0.253, p = 0.006, d.f. = 121, with the z-scores of degraded CPT), whereas nonsignificant correlations between the performances of CPT and the expression of other isoforms. The expression of Esv1 isoform was significantly higher in the CPT deficit subgroup (z-scores of undegraded CPT or degraded CPT < −2.5 as the same definition in the genetic association study) than that in the CPT non-deficit subgroup and control group (Supplementary Table 4). There were no significant differences in the expression of other isoforms among the three groups.

Luciferase reporter assay

The in-vitro functional study of rs11122324 using luciferase reporter assay revealed that the relative transcription activity (ratio to the control vector (pCMV4.10)) of the A-allele vector (pCMV4.10-DISC1-A) was significantly higher than that of the G-allele vector (pCMV4.10-DISC1-G), with an average increase of 20%. The findings were consistent and reproducible in the two cell lines (t = 2.682, d.f. = 5, p = 0.044 for U251 cell line; t = 3.390, d.f. = 3, p = 0.043 for HEK293T cell line) (Fig. 2).

Fig. 2
figure 2

Luciferase reporter assays for different alleles of rs11122324. a Transfection experiments in U251 cell line. Ratio of luciferase activity compared with the control vector (pCMV4.10) in the two vectors with different alleles of rs11122324 of DISC1 3′-UTR (pCMV4.10-DISC1-A and pCMV4.10-DISC1-G). Data are represented as mean ± SE. An asterisk denotes a significant difference (p < 0.05). b Transfection experiments in HEK293T cell line. Data are represented as mean ± SE. An asterisk denotes a significant difference (p < 0.05)

Full size image

Discussion

First, the distribution of genotypes of two SNPs, rs11122324 (at 3′-UTR of Es and Esv1 isoform of DISC1) and rs2793091 (at intron 4), was significantly different between the multiplex schizophrenia patients with deficient sustained attention and the normal controls. The proportions of AA genotypes of the two SNPs were significantly higher in the patient subgroup. The patients with AA genotypes of the two SNPs had poorer CPT performance. Second, the mRNA expressions of Es and Esv1 isoforms were significantly higher in the EBV-transformed lymphoblastic cells of schizophrenia patients than those of controls. The patients with A-allele of rs11122324 had higher expression of Es isoform than those without A-allele. The expression level of Esv1 isoform was negatively correlated to the performance of CPT. Lastly, the vector with A-allele of rs11122324 in 3′-UTR of DISC1 had significantly higher relative luciferase activity than that with G-allele.

Although significant association was not found between DISC1 and schizophrenia as a whole, two SNPs of DISC1 were found to be associated with a specific subgroup of schizophrenia, i.e., multiplex schizophrenia patients with deficient sustained attention. Two-SNP haplotypes composed of rs2793092 (at intron 3) and rs2793091 (at intron 4) were found to have significant transmission distortion, especially in the multiplex schizophrenia families with deficient sustained attention in our previous study [10]. After excluding the overlapping multiplex sample (87 multiplex patients) used in the previous study, the association results remained similar (Supplementary Table 5). Therefore, we independently replicated the association evidence of DISC1 with multiplex schizophrenia patients with deficient sustained attention and the association region was across intron 3 and intron 4.

The two SNPs are not in the same haplotype block according to Haploview [40]. Therefore, the association signals brought up by the two SNPs are likely to be independent. The SNP rs11122324 is located at the 3′-UTR of Es and Esv1 isoforms. The AA genotype was over-represented in multiplex patients with deficient attention. The A-allele correlated with higher mRNA expression of Es isoform in patients and in-vitro function study showed that the A-allele was associated with higher expression activity than the G-allele. Higher expression of Es and Esv1 isoforms were found in schizophrenia patients. All of these consistent findings imply that rs11122324 may have functional influences on the mRNA expression of Es or Esv1 isoform, which contributes to the disease mechanism in the subgroup of multiplex schizophrenia with impaired sustained attention. The genotypes of SNP rs11122324 were reported to be significantly associated with the mRNA expression of total DISC1 in the lymphocytes of healthy individuals and the neighboring SNP rs2487453 (about 4 kb distal from rs11122324) was significantly associated with the thickness of cingulate cortex in healthy individuals [15].

The SNP rs2793091 was also reported to be significantly associated with multiplex schizophrenia families with deficient sustained attention in our previous study [10]. However, the functional relevance of this SNP is still unknown. Clarification is needed about whether the SNP itself is a risk polymorphism or is in linkage disequilibrium with other risk variants.

Another possibility for the association signals brought up by the two SNPs might arise from the functional (rare or low-frequency) variants near the two SNPs. Multiplex schizophrenia patients come from families with heavier genetic loading than simplex schizophrenia patients and have been reported to have more severe impairments in some neurocognitive functions, including sustained attention [41]. Previous studies also suggest that familial schizophrenia (with higher genetic loadings) is the more serious subtype with more minor physical anomalies, more cognitive impairment, and severe structural brain abnormalities compared with sporadic schizophrenia (with lower genetic loadings) [42, 43]. Patients from multiplex families may have more inherited private rare or low-frequency variants than those from simplex families. In the present study, the AA genotypes of the two SNPs were found to be more prevalent in the multiplex patients with deficient sustained attention. If a significant proportion of the patient subgroup with AA genotype have nearby private rare or low-frequency variants, which are inherited in the multiplex families and have more functional impacts on sustained attention, it may be consistent with the association pattern observed in this study, which showed significance only for genotype distribution, but not for allele frequency. Three rare or low-frequency variants were identified within 735 bp proximal to rs11122324 by direct sequencing, two of which are not identified in our controls and have much higher allele frequency compared with those in the public resources (Supplementary Table 2). Rare or low-frequency variants in DISC1 may have roles in the specific subgroup of familial schizophrenia, although we found associations with two common variants.

The original evidence of DISC1 association came from a rare chromosome abnormality in a big Scottish family [4]. Song et al. [44] found 5 ultra-rare missense variants in 6 schizophrenia patients, which were not found in the 288 controls and also were absent in a pool of 10,000 control alleles. The burden analysis was significant. An increased burden of rare missense variants were also reported in exon 11 of DISC1 for schizoaffective disorder [45]. A deep sequencing of 528 kb of DISC1 genomic region in 653 psychiatric patients and 889 controls found 2010 rare variants, 62% of which are not found in 1000 Genomes, and the investigators concluded that many DISC1 SNPs remained undiscovered and are essentially private [46]. Although we sequenced only 24 kb of DISC1 genomic region in only 100 schizophrenia patients, 19% of the identified variants were not found in 1000 Genomes and other public variation resources, which is consistent with the conclusions of that study [46]. However, our sequencing range and sample size were too small to make any conclusion about the roles that rare variants of DISC1 might play in schizophrenia.

We found the association evidence between the two SNPs of DISC1 and the sustained attention of schizophrenia. However, because of the lack of CPT data in the control group in this genetic study, we cannot test the genetic association between DISC1 and the dimension of sustained attention in general. We found significant association between the expression level of Esv1 isoform and sustained attention in both schizophrenia and normal control. Although DISC1 may not be a key risk factor for schizophrenia nor for other disorders that are defined by the DSM diagnostic system, our study suggests that it may confer a genetic risk at the level of sustained attention that underlies several major mental disorders.

In the present study, the mRNA levels of Es and Esv1 isoforms were found to be significantly higher in the lymphoblastic cells of schizophrenia patients. A previous postmortem brain study reported a significant increase of mRNA expression of Esv1 isoform in the hippocampus of schizophrenia patients [27]. The function of the Es/Esv1 isoforms is still unknown, although enriched expression was reported in the fetal brain [27]. The functional relevance of increased level of Es and Esv1 isoform may be associated with schizophrenia in a dominant-negative manner, because the truncated proteins translated by the ultrashort isoforms lack the C terminus predicted to mediate protein–protein interactions with a large number of proteins. Transfection of C-terminally truncated DISC1 into cells leads to aberrant phenotypes, such as disruption of interactions with other proteins, including NDEL1 [47], MAP1A, MIPT3, ATF4/5 [48], and PDE4B [23], aberrant mitochondrial function [49], and inhibition of neurite outgrowth [20, 47]. However, the definite effects of overexpression of these ultrashort isoforms for cells and animals still remain to be clarified.

This study has several limitations. First, the association evidence of the two SNPs was limited to the genotype distribution, but not allele frequency. Second, although patients with the A-allele had higher expression of Es isoform, the association between genotype and expression is not compatible with the recessive model in the genetic association. Third, the higher expression of mRNAs of Es and Esv1 isoform between patients and controls is not significant after multiple correction.

Findings of this study suggest that the SNP rs11122324 at the 3′-UTR of Es and Esv1 isoforms may be a true risk polymorphism through its influence on the expression of extra-short isoforms. Alternatively, some true functional variants may be harbored near the two SNPs in the specific familial schizophrenia subgroup. Deep sequencing for the specific subgroup is worth performing. DISC1 may also play roles on the neural mechanism of sustained attention in schizophrenia. The functional impact of elevated extra-short isoforms of DISC1 upon cell and animal models is also worth further exploration.