Schizophrenia is a devastating mental disorder with a significant genetic component that affects approximately 0.5–1.0% of the general population. Based on genetic epidemiological studies, it was estimated that schizophrenia has a heritability of 60-80%[1, 2]. Disrupted-in-schizophrenia 1 (DISC1) is one of the candidate susceptibility genes for a spectrum of major psychiatric disorders. DISC1 was originally identified on chromosome 1 by analyzing a large Scottish pedigree showing a heavy burden of major psychiatric disorders associated with balanced chromosomal translocation (1:11)(q42.1:q14.3)[3, 4]. DISC1 is implicated in the genetic risk for many aspects of quantitative cognitive traits of psychotic patients[5–8]. Accumulating evidence indicates that DISC1 acts as a scaffold protein that in concert with numerous interacting proteins regulates neurogenesis, neuronal migration, neurite outgrowth, and signal transduction[9–11].

Mouse models for the genetic mutation of Disc1 are of relevance for unraveling the impact of gene disruption on neural integrity associated with schizophrenia. A number of conditional transgenic mouse models overexpressing truncated versions of human DISC1 protein under various exogenous promoters have been reported. In these models, truncated versions of human DISC1 protein are overexpressed either constitutively[12–14] or transiently during early postnatal life[15–18]. A mouse strain carrying a truncating lesion (25-bp deletion in exon 6) in the endogenous Disc1 orthologue has also been reported[19–21]. Although these mouse models display a range of psychiatric disease-related behavioral phenotypes, many have been inconsistent across studies due to differences in behavior tests and experimental design and the specific genetic aberrations.

An alternative method for generation of a mouse Disc1 model is to screen an archive of G1 mice carrying many N-ethyl-N-nitrosourea (ENU)-induced mutations. With this approach, it is feasible to identify an allelic series of ENU-induced mutations in a particular gene, in this case Disc1, as a gene-driven approach based on a priori assumptions of the function of a gene of interest[22, 23]. Genomic DNA libraries from thousands of G1 mice born to ENU-mutagenized C57BL/6 J male mice were screened for a point mutation in Disc1, one of which was subsequently identified as conferring a L100P amino acid substitution in exon 2[24]. Roder and his collaborators reported that the Disc1 L100P mouse line showed schizophrenia-like behavior, such as increased basal and amphetamine-induced open field activity, and deficits in prepulse inhibition (PPI)[24, 25]. However, other independent research groups observed neither an increase in basal locomotor activity nor abnormal PPI[26, 27].

Because ENU generates multitude, randomly distributed, single-base point mutations throughout the entire mouse genome, ENU mutant mouse lines may harbor unknown functional mutations other than those are assumed to contribute to the phenovariance of interest[22, 23, 28, 29]. Additionally, ENU-mutagenized C57BL/6 J male mice at RIKEN were out-crossed to untreated DBA/2 J female mice to produce the original progeny (G1). Although heterozygous mice resulting from repeated backcrossing to the C57BL/6 J background were finally intercrossed to generate L100P/L100P homozygous mice, remaining ENU-induced mutations outside Disc1 and/or single nucleotide polymorphisms (SNPs) from DBA/2 J background may confound the causal relationship between L100P mutation and the behavioral phenotype. Disclosure of this information would provide a justification for the usage of ENU-mutagenesis to fully explain the phenotypic manifestation of the main causative mutation. Herein, we performed three experiments to better understand behavioral phenotypes of Disc1 L100P mice with reference to our new genetic data. First, we investigated the genetic background of the L100P strain that was reversely-deposited back to the Experimental Animal Division (EAD), RIKEN BioResource Center (BRC) by Roder’s group[24]. The genetic background of the deposited L100P/L100P homozygous pairs and their offspring born via intercrossing between the homozygotes were investigated. Second, L100P/L100P, L100P/+ and wild-type littermates (+/+), which were derived from the deposited L100P/L100P homozygous mice mated with the C57BL/6 J mice (details in Materials and Methods), were examined for schizophrenia-related behavioral phenotypes using open field and social interaction tests. Third, we conducted whole-exome resequencing analysis of the original G1 genome to screen for ENU-induced mutations in addition to the Disc1 L100P mutation. The results obtained were discussed in light of the validity and issue of Disc1 L100P mutant model.

The present study compared schizophrenia-related behavior among L100P mutant mice (L100P/L100P and L100P/+) and their wild-type littermates, and inbred C57BL/6 J mice by testing spontaneous locomotor activity, methamphetamine-induced locomotor activity in the open field[33], and sociability/social novelty preference in the social interaction[34]. All of behavior was comparable among L100P/L100P, L100P/+ and wild-type littermates; the results were partially inconsistent with previous studies using mice originated from the same G1 founder[24, 25]. To assess effects of genetic background (C57BL/6 J vs DBA/2 J), we conducted comprehensive genotyping of the original G1 mouse and its homozygous offspring obtained by backcrossing to C57BL/6 J, the same strain used for backcrossing in the previous report characterizing the L100P mutant[24]. Our genetic analyses demonstrated that both the L100P/L100P homozygous mice deposited by Roder’s group[24] into the EAD at RIKEN and their homozygous progeny still harbored a small number of SNPs inherited from the DBA/2 J strain (the mother of the G1 founder). Whole-exome resequencing of the original G1 genome also revealed an additional 116 single nucleotide variants induced by ENU.

Disc1 L100P was originally identified in one G1 progeny of several thousand screened for Disc1 mutants from a G1 male genomic DNA archive produced by breeding ENU-mutagenized C57BL/6 J males and untreated DBA/2JJcl females[24]. Live heterozygous mice carrying the L100P mutation were recovered by in vitro fertilization of C57BL/6 J eggs with the cryopreserved sperm of the corresponding G1 progeny. In the study reported by Clapcote et al.[24], mutant mice were then backcrossed to the C57BL/6 J background for at least six generations before intercrossing L100P/+ mice to generate homozygous mutants for behavioral tests. In theory, the backcrossed congenic mutant strain should have lost most of the other ENU-induced mutations and SNPs derived from DBA/2 J; however, the rate of SNP and mutant loss per generation has not been examined experimentally.

DBA/2 J and C57BL/6 J inbred strains were shown to exhibit differential sensitivity to psychostimulants[35, 36] without significant differences in basal locomotor activity or sociability[31]. Thus, residual DBA/2 J SNPs, if any still exist in the analyzed mice, could conceivably influence the behavioral test results. Genomic analysis revealed a significant number of DBA/2 J SNPs remaining in the backcrossed Disc1 L100P mice. Among 117 loci tested, nineteen were still polymorphic in the eight L100P/L100P mice deposited at RIKEN and their progeny (Additional file2: Table S1). The average frequency of DBA/2 J SNP alleles in the eight L100P/L100P mice was 6.63% (5.56%–8.12%, Additional file2: Table S1); a rate of genetic vestige was 4.25-fold higher than the theoretical estimate of 1.56%[24]. The question arises as to whether any single DBA/2 J SNP or combination influences Disc1 L100P locomotor activity in the open field compared with the congenic background strain. An important feature of the present study was the inclusion of commercially available inbred C57BL/6 J mice in addition to wild-type littermates derived from Disc1 L100P mutants to assess effects of genetic background on behavioral phenotype. This experimental design could enhance the sensitivity for detecting effects of genetic background and ENU-induced mutations. However, our behavioral analyses showed that neither basal nor methamphetamine-induced locomotor activity differed between wild-type mice derived from Disc1 L100P mutants and inbred C57BL/6 J mice. This finding suggests that locomotor activity is not measurably affected by residual DBA/2 J SNPs. Subsequently, using the exome resequencing analysis, we identified 116 previously unreported mutations in the exome of the original G1 male, although L100P was the only single mutation found in the coding sequence of the Disc1 gene. It has been estimated that approximately 64%, 26% and 10% of ENU-induced single base pair mutations in coding regions are missense, synonymous, and nonsense mutations, respectively[28, 37, 38]. Intriguingly, the frequencies of base substitutions in the G1 exome decreased in the rank order of A/T to G/C transition > AT/TA transversion > G/C to A/T transition > G/C to T/A transversion > A/T to C/G transversion, consistent with estimates for the gene-driven approach[39]. Furthermore, the distribution of single base pair mutations in the G1 exome predicted to generate amino acid variants was approximately consistent with previous studies[37, 38].

Locomotor activity in the open field did not differ between L100P mutant mice (homozygotes and heterozygotes) and wild-type littermates, consistent with the report by Shoji et al.[27]. In contrast, Clapcote et al.[24] and Lipina et al.[25] reported higher locomotor activity in L100P/L100P mice than in wild-type littermates during the first 30 min, after introduction into the open field (but not thereafter). The higher locomotor activity of L100P mice may represent slower habituation to a novel environment. In addition, although Lipina et al.[25] reported higher locomotor activity in L100P mice following injection of amphetamine compared with that of wild-type littermates, we did not observe enhanced locomotion after an acute challenge with methamphetamine. Although the reasons for this discrepancy with previous reports remains to be explored, one possibility is that critical mutation(s) other than L100P may have been missed during maintenance of the strain after submission to the RIKEN EAD and/or the single backcrossing of homozygous (L100P/L100P) mutant males obtained from RIKEN to female inbred C57BL/6 J mice. As shown by the distribution of SNPs, the genetic background in the deposited L100P homozygotes was heterozygous; therefore, such heterogeneity may be fixed to either allele in the offspring, even after many generations of maintenance.

The enhanced release of amphetamine-evoked dopamine consistently reported in neuroimaging studies of schizophrenia[40–42] may be of relevance to this model of schizophrenia. However, Lipina et al.[25] did not find greater striatal release of dopamine following amphetamine injection in L100P mice. In the adult brain, Disc1 is expressed in the hippocampus and not in the nucleus accumbens[43, 44]. The hyperactivity of subcortical dopaminergic neurons is believed to be conveyed by glutamatergic afferents from the hippocampus to the nucleus accumbens, which in turn regulates release of dopamine and the activity of dopaminergic neurons in the ventral tegmental area[45, 46]. Another possible mechanism relates to disruption of the interaction between the hippocampus and prefrontal cortex[47]. Although subtle cytoarchitectonic changes have been reported in frontal cortical neurons[48], Disc1 L100P mice lack the reduced number of parvalbumin-positive interneurons observed in Disc1 transgenic models[12, 18].

Sociability and social novelty preference tests used in the present study are used frequently to investigate disrupted social interactions in mouse models of schizophrenia[34]. The impoverished social interaction in Disc1 mice is believed to be a relevant quantitative trait of social withdrawal in schizophrenia[14]. Consistent with Clapcote et al.[24], we found that the three L100P genotypes did not differ in sociability or social novelty preference. Intriguingly, sociability was markedly impaired in conditional transgenic lines with either constitutive expression of[14] or inducible expression of a mutant DISC1 C-terminal fragment during the early postnatal period[15]. We hypothesize that transgenic mice overexpressing human truncated DISC1 protein exhibit a dominant-negative effect that may lead to diminished binding between the DISC1 interacting proteins and relevant domains of DISC1 protein, thus altering the behavioral phenotype[15]. Based on these results, the relevance of Disc1 L100P as a mouse model of schizophrenia should be re-evaluated.

A recent review stressed that ENU-mutagenized mice are useful for establishing novel models of complex human diseases, including neuropsychiatric diseases[49]. A range of ENU-generated mutations based on a phenotype-driven approach has an advantage over a complete loss-of-function mutation in recapturing the pleiotropic nature of human neuropsychiatric diseases and the subtlety of manifestation[49]. Thus, both disease-causative and other unexpected variants should be made publicly available[28, 29]. This phenotype-driven approach requires high-throughput behavioral screening to link mutation with function. One promising candidate screening methodology is the identification of outliers by neuroimaging. For example, only a few studies on developmental mouse models of psychiatric illness, including those focusing on DN-DISC1 models, have highlighted enlargement of the lateral ventricle without a marked change in overall brain size[16–18, 50]. Phenotype screening based on standardized behavioral tests[31, 51] combined with neuroimaging[50] to detect enlargement of the lateral ventricle may be an efficacious approach for the identification of robust schizophrenia models.

Several limitations of this study should be considered. First, we did not conduct a mouse comprehensive battery of behavioral tests including cognitive tasks and PPI but only tests of locomotor activity in a novel environment, psychostimulant-induced behavior and sociability in the three-chamber test. Therefore, our results do not clarify effects of the L100P amino acid substitution in exon 2 on other behavioral phenotypes related to psychiatric disorders including schizophrenia. Second, handling, experimental protocols, rearing environment and other environmental factors may influence behavioral test results and contribute to variability across studies[51, 52]. Third, L100P mice used in the current study were not genotyped for all ENU-induced mutations found in the exome and vicinity of the G1 genome, some of which may have been transmitted to progeny despite numerous backcrosses[24]. Given that 117 ENU-induced mutations in the G1 mouse identified with the whole-exome resequencing, it is plausible that 50-fold excess or more than 5,000 of ENU-induced mutations should exist in the entire G1 genome. Although detailed genetic information on models relevant to human disease should be freely available, routine implementation of whole-genome sequencing awaits measures for improved cost-effectiveness. However, as most of the ENU-induced mutations in G1 map to chromosomes other than chromosome 8 (containing Disc1), it is likely that they are transmitted to progeny to some extent regardless of the breeding process used to select mice harboring Disc1 100P. Although it may not be possible to investigate whether genetic background or other ENU-induced mutations have any effects on behavior because of a generalised lack of information on genotype differences among any of their mouse lines or offspring, further behavioral and genetic studies are warranted. For example, generation and breeding of another mouse line harboring Disc1 L100P, which can be accomplished by in vitro fertilization by the cryopreserved sperm of the G1 progeny owned by RIKEN BRC, may facilitate full elucidation of a relationship between genetic components, including Disc1 L100P mutation per se, and behavioral phenotypes, related with schizophrenia in this mouse line.

The present study using behavioral genetic approaches provides an insight into the role of Disc1 L100P and other single nucleotide variants in behavioral phenotypes associated with psychiatric disorders such as schizophrenia. Our present findings suggest that causal attribution of the discrepancy in behavioral phenotypes of the Disc1 L100P mutant mouse line existing among different research groups, including our own, needs to be cautiously investigated in further studies by taking into account the effect(s) of other ENU-induced mutations and/or SNPs from DBA/2 J. Further behavioral genetic analyses are needed to elucidate the cause of behavioral variance associated with Disc1 L100P strain. Using a polygenic animal model including ENU mutagenesis provides an efficacious approach to explore the relationship of variants to behavioral phenotypes associated with polygenic and multifactorial disorders such as psychiatric disorders.