Introduction

Schizophrenia (SCZ) is a chronic and severe mental disorder characterized by some combination of hallucinations, delusions, and extremely disordered thinking and behavior that impairs daily functioning [1]. The lifetime prevalence of SCZ is 0.3–0.7%, and the standardized mortality ratio is 2.5 [2, 3]. Both genetic and environmental factors affect the risk for SCZ [4, 5]. Population-based and twin concordance studies indicated that the heritability of SCZ is 60–80%, with only subtle contributions from environmental factors [6, 7].

As genetic factors play a significant role in the etiology of SCZ, many genetic studies of the disease have been conducted. Linkage studies identified multiple genetic risk loci for SCZ; however, these studies were not sufficient to suggest specific causative genes [8]. Genome-wide association studies (GWASs) have identified numerous loci significantly associated with SCZ [9, 10] based on the common disease–common variant hypothesis. However, the effect sizes of the individual single-nucleotide polymorphisms identified in GWASs are too small to explain the high heritability of SCZ demonstrated in cohort studies, which has been designated “missing heritability” [11]. To overcome this problem, it is necessary to consider rare variants such as single-nucleotide variants (SNVs) or copy number variants [12].

The feasibility of exploring disease-associated SNVs has been enhanced due to the advent of whole-genome sequencing (WGS) and whole-exome sequencing (WES) techniques, with the latter providing a more rapid and cost-effective approach for sequencing protein-coding regions across the genome. However, efficiently extracting candidates from a large number of detected mutations is challenging. As such, most of these studies have focused on de novo genetic variants or rare variants significantly associated with SCZ due to interpretability [13–16] in terms of their large estimated effect sizes and the possibility of functional validation.

In addition, patients in families with multiple affected members are likely to be enriched in genetic factors that strongly affect the development of SCZ [17]. Therefore, it is also essential to investigate variants shared among patients in order to elucidate the association between transmitted variants and SCZ. Indeed, several sequencing studies have focused on shared variants associated with bipolar disorder [18, 19] and autism spectrum disorder [20, 21] in patients in the same multiplex family. A previous WES study of SCZ multiplex families demonstrated that several loci could potentially affect synaptic plasticity and neurocognitive performance [22] and that variants in genes related to metabotropic glutamate receptor 5 (mGlu5) are more common in affected family members [23].

However, the above-mentioned relatively large-scale sequencing studies have been done in Caucasian populations. The total genetic variation of the Japanese population, however, is considered relatively low [24] compared to that of ethnically diverse populations, in particular Europeans, which can be beneficial in sequencing studies due to decreased allelic diversity [25–27]. Furthermore, rare variants, which were targeted in this study due to their more recent origin, tend to be more geographically clustered and can be population specific, thus potentially revealing new SCZ candidates beyond those identified in studies of Caucasian populations.

We therefore hypothesized that we could identify variants associated with SCZ susceptibility in the Japanese population via analyses of Japanese SCZ multiplex families, even if the sample sizes were small compared with studies of Caucasian populations. Individual rare variants provide limited power for identifying significant trait associations, and thus, multi-variant and/or multi-genic approaches such as gene set enrichment tests are necessary.

To address this hypothesis, a WES study of multiplex families within the Japanese population was conducted to identify disease-associated rare variants or gene sets.

Materials and methods

Participants—SCZ multiplex families

A multiplex family was defined as a family having more than one member with SCZ. DNA samples were collected from peripheral blood or saliva of 29 patients with SCZ, 1 patient with obsessive-compulsive disorder (OCD), and 9 healthy individuals from 14 SCZ multiplex families in Japan (Fig 1, S1 Table). Among the 14 SCZ multiplex families, pedigree 10 was a consanguineous family. All families were unrelated, lived on the mainland of Japan, and self-identified as Japanese. All patients fulfilled the criteria for SCZ listed in the Diagnostic and Statistical Manual of Mental Disorders–Fifth Edition (DSM-5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Sequenced samples.

Dashed lines indicate individuals who underwent WES. Squares indicate males, and circles indicate females. Shading indicates an affected individual, and slash marks indicate deceased individuals. We performed WES for 29 patients with SCZ, 1 patient with obsessive-compulsive disorder (OCD), and 9 healthy individuals from 14 SCZ multiplex families in Japan.

https://doi.org/10.1371/journal.pone.0268321.g001

The Ethics Committees of Nagoya University Graduate School of Medicine approved this study, and written informed consent was obtained from all subjects. The study was conducted in accordance with the Helsinki Declaration of 1975 and its later amendments or comparable ethical standards.

Results

WES analysis

We performed WES for 14 SCZ multiplex families in Japan (Fig 1). The data were quality checked and filtered using the following criteria: 1) low frequency (an allele frequency ≤1% in the databases), 2) protein-altering (missense and splice site variants), and 3) variants in the intolerant genes (RVIS ≤ 25%) and genes expressed in brain regions. Finally, we identified 525 variants among 481 genes carried by at least two patients in the same family. A summary of the filtered variants is presented in S2 and S3 Tables. Loss-of-function variants among these candidate variants are presented in Table 1. In addition to the 525 variants identified among the 481 genes, we also identified a number of de novo variants (Table 2). Therefore, we identified a total of 530 variants among 486 genes as potential multiplex family candidate variants for SCZ.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Loss-of-function variants.

https://doi.org/10.1371/journal.pone.0268321.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Homozygous (hemizygous) and de novo variants.

https://doi.org/10.1371/journal.pone.0268321.t002

Of the 14 pedigrees examined, seven (pedigrees 1, 4, 5, 10, 11, 13, and 14) included control samples. We selected 199 variants in 190 genes found in these seven pedigrees as “strict-filtered” for further analysis, as we sequenced both affected and unaffected family members in these pedigrees to exclude variants carried by healthy individuals.

GO analysis

To identify gene categories that could be affected by the identified candidate variants, we performed GO analysis. Significantly enriched categories (Q-value <0.05) are shown in Table 3. We tested the enrichment of the 199 variants identified in the 190 “strict-filtered” genes (pedigrees 1, 4, 5, 10, 11, 13, and 14) and found significant enrichment of genes associated with calcium channel activity (Q = 0.032). A similar result was also observed for the 530 variants identified in 486 genes that were derived from whole pedigrees in this study (Q = 0.011).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. GO analysis results.

https://doi.org/10.1371/journal.pone.0268321.t003

Discussion

This is the first report of a WES analysis of Japanese SCZ multiplex families. After WES analysis and filtering of 14 SCZ multiplex families, we selected variants shared only among patients (we also selected de novo variants only from pedigrees whose parents were unaffected, as pedigrees with affected parents could have shared rare variants between cases), and we identified a total of 530 SCZ candidate SNVs and indels among 486 genes. In an in silico analysis involving 530 SCZ candidate SNVs and indels among 486 genes, we demonstrated that many candidate variants were located in genes related to calcium ion channels that have also been reported as involved in the pathophysiology of SCZ, as demonstrated by common variants from genome-wide association analyses [10] and rare variants from whole-genome CNV/SNV analyses [16, 36]. For example, CACNA1C, which was identified as a candidate gene in this study, was also identified as a SCZ susceptibility gene in a GWAS [10] as well as exome [16] and whole-genome CNV [36] analyses.

The 486 candidate genes from 530 SCZ candidate variants identified in our present study were enriched in calcium channel activity–related genes as the most enriched GO terms (Table 3), although there could have been enrichment of calcium channel–related GO terms among the top 25% RVISs and brain-expressed genes. Voltage-gated calcium channels are widely distributed in all parts of the brain. They are critical for mediating intracellular Ca²⁺ influx, which results in transmitter release from pre-synaptic endings, thereby affecting neuronal excitability and synaptic plasticity and playing a role in neurodevelopmental disorders such as SCZ [37]. On the other hand, in this study, we did not observe any GO enrichment in the SCZ-associated genes reported in two previous WES studies of SCZ multiplex families [22, 23] such as AMACR, a gene involved in fatty acid metabolism and previously implicated in SCZ [22], and genes related to the metabotropic glutamate receptor 5 (Table 3) [23].

Among the 530 candidate variants identified, we detected a male carrier of a hemizygous variant in MAOB (located on the X chromosome) and one homozygous variant in CNTN6 in a pedigree with consanguineous marriage (Table 2). The p.P109L variant in MAOB identified in family 13 was previously reported in a male SCZ patient as being inherited from his heterozygous mother [38]. This variant may cause a change in the structure of the protein’s binding site to the mitochondrial membrane [39]. We also identified this hemizygous variant in a male SCZ patient segregated from his heterozygous mother with OCD (Fig 1). Interestingly, MAOB is suggested as being associated with OCD [40].

The homozygous SNV in CNTN6 shared among affected members of pedigree 10 with consanguineous marriage is also interesting, as several recent reports have identified CNTN6 as a candidate gene involved in neurodevelopmental disorders, including SCZ [41, 42]. SNVs in CNTN6 are significantly associated with Autism spectrum disorder in particular [43]. CNTN6 encodes contactin 6, which plays a role in neuronal cell adhesion and promotes neurite outgrowth in sensory-motor neuronal pathways [44]. The homozygous variant we identified is located in the immunoglobulin C2–type domain (IGC domain), which mediates interactions with contactin-binding partners such as protein tyrosine phosphatase receptor–gamma (Ptptg), which plays a role in the molecular basis of neurodevelopmental functions [44]. Moreover, among the 486 genes examined in the present study, three genes (RBM12, NRXN1, AKT1) are registered in the OMIM (https://www.ncbi.nlm.nih.gov/omim) database as being associated with susceptibility to SCZ.

Although our results provide some support for the calcium channel–associated hypothesis, there are still limitations to our study. First, our sample size was small for an investigation of the burden and/or transmission of rare and de novo protein-altering mutations between affected and non-affected individuals, and thus, our results could be misleading. Second, we could not evaluate clinical phenotypes, especially within unaffected families. Therefore, we could not evaluate the clinical phenotype of unaffected family members with discovered variants associated with susceptibility to SCZ, such as MAOB and CNTN6. Furthermore, in this study, we could only elucidate minimal variant information for each pedigree due to data availability constraints. In future multiplex familial studies, it will be important to evaluate in detail the relationship between variants and phenotype within the same family. Third, due to the study’s small sample size, the experiments could have generated false-positive results, and strict filtering could have led to some false-negative results. The different strategies we used in the study to narrow down candidate variants likely limited false positives, however. More importantly, SCZ is not a monogenic disorder, so filtering is not necessary to identify only a few pathogenic variants; identifying disease-associated pathways would also be a useful approach to elucidate the mechanism of the disorder.

Conclusion

In conclusion, using WES analysis of 14 Japanese multiplex families, we identified a number of rare variants segregated in SCZ patients. Our results provide support for the hypothesis that calcium channel activity is related to the development of SCZ. Analyzing a larger sample size of multiplex families could confirm these results and provide additional information regarding the aspects of this disease.

Supporting information

Acknowledgments

We are grateful to all of the patients and their families who contributed to this study. We thank Mami Yoshida, Kiyori Monta, Hiromi Noma, and Yukari Mitsui for technical assistance, discussions, and contributions to creating and managing the database.