Role of an atypical cadherin gene, Cdh23 in prepulse inhibition and its implication in schizophrenia

TL;DR: In this paper, a quantitative trait loci (QTL) for prepulse inhibition (PPI), an endophenotype of schizophrenia, was identified on mouse chromosome 10 and reported Fabp7 as a candidate gene from an analysis of F2 mice from inbred strains with high (C57BL/6N; B6) and low (C3H/HeN; C3H) PPI levels.

Abstract: We previously identified quantitative trait loci (QTL) for prepulse inhibition (PPI), an endophenotype of schizophrenia, on mouse chromosome 10 and reported Fabp7 as a candidate gene from an analysis of F2 mice from inbred strains with high (C57BL/6N; B6) and low (C3H/HeN; C3H) PPI levels. Here, we reanalyzed the previously reported QTLs with increased marker density. The highest LOD score (26.66) peaked at a synonymous coding and splice-site variant, c.753G>A (rs257098870), in the Cdh23 gene on chromosome 10; the c.753G (C3H) allele showed a PPI-lowering effect. Bayesian multiple QTL mapping also supported the same variant with a posterior probability of 1. Thus, we engineered the c.753G (C3H) allele into the B6 genetic background, which led to dampened PPI. We also revealed an e-QTL (expression-QTL) effect imparted by the c.753G>A variant for the Cdh23 expression in the brain. In a human study, a homologous variant (c.753G>A; rs769896655) in CDH23 showed a nominally significant enrichment in individuals with schizophrenia. We also identified multiple potentially deleterious CDH23 variants in individuals with schizophrenia. Collectively, the present study reveals a PPI-regulating Cdh23 variant and a possible contribution of CDH23 to schizophrenia susceptibility.

Summary

Introduction

• Deciphering neurobiological correlates for behavioral phenotypes in terms of genetic predispositions that affect molecular, cellular and circuit-level functions is crucial for understanding the pathogenesis of psychiatric disorders.
• Polygenicity and the inherent limitations of psychiatric diagnosis based on the subjective experiences of patients have impeded this endeavor.
• 6, 7 As a robust and heritable endophenotype in schizophrenia, 8-11 the genes/variants conferring the risk of dampened PPI can help to elucidate the genetic architecture of schizophrenia.
• 12, 13 We have previously performed a large-scale quantitative trait loci (QTL) analysis for PPI and mapped six major loci through an analysis of 1,010 F2 mice derived by crossing selected inbred mouse strains with high (C57BL/6NCrlCrlj: B6Nj) and low (C3H/HeNCrlCrlj: C3HNj) PPI.the authors.the authors.
• 14, 15 However, this high LOD score cannot be explained by a single gene, considering the low phenotypic variance attributed to the individual genes/markers, suggesting the presence of additional causative gene(s).

Animals

• All animal experiments were approved by the Animal Ethics Committee at RIKEN.
• For the QTL analysis, the authors genotyped F2 mice from their previous study.
• 14 For other experiments, C57BL/6NCrl (B6N) and C3H/HeNCrl (C3HN) mice were used (Japan's Charles River Laboratories, Yokohama, Japan).
• Note that the previously used mouse strains C57BL/6NCrlCrlj (B6Nj) and C3H/HeNCrlCrlj (C3HNj) are no longer available because their supply ended in 2014.

Human DNA Samples

• Studies involving human subjects were approved by the Human Ethics Committee at RIKEN.
• All participants in the genetic studies gave informed written consent.
• 17, 18 Additionally, the authors also used genetic data of 8,380 healthy Japanese controls from the Integrative Japanese Genome Variation Database by Tohoku Medical Megabank Organization .

Human induced Pluripotent Stem Cells (hiPSCs)

• To test the CDH23 expression during neurodevelopment in vitro, hiPSC lines established from healthy controls (n = 4) were used.
• 19 To analyze allele-specific expression of CDH23 for the rs769896655 (c.753G>A) variant, another hiPSC line was established from a healthy individual (TKUR120) who was heterozygous for the variant.
• Establishment of hiPSCs and their differentiation to neurospheres and neurons were performed as described before.

QTL Analysis

• Reanalysis of PPI-QTL was performed by increasing the density of the analyzed markers.
• 14 The 148 additionally selected SNV markers (supplementary table 1) were genotyped in 1,012 F2 mice by Illumina BeadArray genotyping (Illumina Golden Gate assay) on the BeadXpress platform as per the manufacturer’s instructions and were added to the analysis.
• QTL analysis was done by a composite interval mapping method 21 using Windows QTL Cartographer v2.5 (https://brcwebportal.cos.ncsu.edu/qtlcart/WQTLCart.htm), and by Bayesian multiple QTL mapping (supplementary methods) .
• It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in 7.

Analyses of PPI and Auditory Brainstem Response

• The PPI test was performed according to previously published methods.
• 14, 15, 25 Auditory brainstem response (ABR) was recorded as described previously (supplementary methods).

Gene Expression Analysis

• To detect the localization of Cdh23 expression in the brain, the authors performed RNA in situ hybridization in mice (B6N and C3HN) and marmoset as described elsewhere (https://geneatlas.brainminds.riken.jp/).
• A variant were tested by semiquantitative RT-PCR normalized to Gapdh (supplementary table 3).
• It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in 8.

Mutation Analysis of CDH23 in Schizophrenia by Targeted Next-generation Sequencing

• (NGS) using Molecular Inversion Probes (MIP) Coding exons and the flanking exon-intron boundaries of the CDH23 were sequenced using MIP as described previously (supplementary table 4).
• 32, 33 Library preparation, sequencing and variant analysis were performed as described in the supplementary methods.

Statistical Analysis

• The data in the figures are represented as the mean ± SEM.
• Statistical analysis and graphical visualization was performed using GraphPad Prism 6 (GraphPad Software).
• The total sample size (n) and the analysis methods are described in the respective figure legends.
• It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in 9.

Cdh23 is Predominantly Expressed in the Subthalamic and Pontine Regions of the Brain

• Cdh23, an atypical cadherin, is a member of the cadherin superfamily, which encodes calcium-dependent cell-cell adhesion glycoproteins.
• It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in 10 and pontine regions in both B6N and C3HN mouse strains across the developmental stages .
• Furthermore, a similar expression pattern was observed in the brain regions from the marmoset (Callithrix jacchus) , indicating functional conservation across species.
• 19, 20 These lines of evidence indicate potential roles for Cdh23 in neuronal function and PPI.

Strains

• Since a high LOD score was found for the SNV marker c.753G>A (rs257098870) in Cdh23, the authors reasoned that this variant or other flanking functional variants in the Cdh23 gene might be responsible for the observed phenotypic effect.
• The c.753A allele was previously known to cause age-related hearing loss (ARHL) by affecting cochlear stereocilia architecture 28, 42, 43 .
• The authors also queried Cdh23 genetic variations in C57BL/6NJ and C3H/HeJ mouse strains from Mouse Genomes Project (MGP) data (https://www.sanger.ac.uk/science/data/mouse-genomes-project/), which again showed c.753G> CC-BY-NC-ND 4.0 International licenseperpetuity.
• It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in 11 c.753G might contribute to the dampened PPI, and the authors set out to evaluate its role in PPI function.

C3HN Strain-specific Cdh23 c.753G Allele Knock-in Mice Showed Reduced PPI Levels

• A variant in modulating PPI, the authors generated a mouse model in the B6N genetic background in which the C3HN strain-specific Cdh23 c.753G allele was knocked in using CRISPR/Cas9n-mediated genome engineering .
• The authors tested the splicing pattern of Cdh23 exon 9 in the brain regions (subthalamic and pontine regions) where Cdh23 was predominantly expressed, from knock-in mice homozygous for the G allele, control littermates homozygous for the A allele, and inbred B6N and C3HN mice.
• The G allele knock-in mice showed lower PPI levels than the A allele mice in all the prepulse levels tested .

Potential Role of the Cdh23 c.753G>A Variant in Cdh23 Transcript Expression

• A variant, the authors examined the transcript expression levels of Cdh23 in pontine and subthalamic regions using digital PCR.
• The authors observed significantly lower Cdh23 transcript levels in the G allele carriers in both brain regions, indicating that c.753G>.
• This trend was not observed in inbred adult B6N and C3HN mice (4-6 weeks-old) .
• Interestingly, differences in Cdh23 transcript expression, between inbred B6N and C3HN mice, was observed during earlier stages of development particularly in E16.5 and P0 .

Role of CDH23 in Schizophrenia

• Mounting evidence from genetic studies has indicated the potential role of CDH23 in schizophrenia and other neuropsychiatric disorders.
• Several patient-specific novel variants with potential deleterious effects predicted by in silico tools were identified in CDH23, although each variant was observed in 1-2 cases only .
• The pooled analysis of schizophrenia samples [Japanese + Schizophrenia exome meta-analysis consortium, ; https://schema.broadinstitute.org/] and all controls, did not show any allelic association (P = .08) (supplementary table 9), which could be attributed to the rarity of variant allele A in other populations compared to the Japanese population.
• Allele-specific expression of the variant rs769896655 in the CDH23 transcript was tested in hair follicles, peripheral blood samples, hiPSCs, hiPSC-derived neurospheres and neurons from a healthy subject who was heterozygous for the variant .

Discussion

• By reanalyzing and fine-mapping mouse chromosome-10 PPI-QTL, 14 the authors revealed Cdh23 as a candidate for the PPI endophenotype.
• The rarity of the variant A allele in humans, the lack of homozygotes and the population specificity indicate that the homologous variant might have originated recently.
• In summary, the current study demonstrates the role of an atypical cadherin gene, Cdh23, in regulating PPI through a genetic variant without affecting hearing acuity, and a potential role of CDH23 in schizophrenia.

Figure legends

• Fig. 1. QTL mapping for PPI and expression analysis of Cdh23 (A) LOD score calculated by composite interval mapping for prepulse level of 86 dB (B) Lod score distribution on chromosome 10.
• (C) Bayesian multiple QTL mapping for prepulse level of 86 dB.
• (G) Allele-specific expression of CDH23 in hair follicles (n = 1 in triplicate), peripheral blood samples (n = 1 in triplicate), hiPSCs (four lines, triplicate measurements), hiPSC-derived neurospheres, and neurons (early neurons; day 7, mature neuron; day 30 of differentiation) from a healthy subject heterozygous for the variant.

Full text

1
Role of an Atypical Cadherin Gene, Cdh23 in Prepulse Inhibition and
Implication of CDH23 in Schizophrenia
Shabeesh Balan
1
, Tetsuo Ohnishi
1
, Akiko Watanabe
1
, Hisako Ohba
1
, Yoshimi Iwayama
1
,
Manabu Toyoshima
1
, Tomonori Hara
1
,
Yasuko Hisano
1
, Yuki Miyasaka
2,3
, Tomoko Toyota
1
,
Chie Shimamoto-Mitsuyama
1
, Motoko Maekawa
1,4
, Shusuke Numata
5
, Tetsuro Ohmori
5
,
Tomomi Shimogori
6
, Yoshiaki Kikkawa
2
, Takeshi Hayashi
7
, Takeo Yoshikawa*,
1
1 Laboratory for Molecular Psychiatry, RIKEN Center for Brain Science, Wako, Saitama,
Japan
2 Deafness Project, Tokyo Metropolitan Institute of Medical Science, Setagaya-ku, Tokyo,
Japan
3 Division of Experimental Animals, Graduate School of Medicine, Nagoya University,
Nagoya, Japan
4 Department of Biological Science, Graduate School of Humanities and Science,
Ochanomizu University, Tokyo, Japan
5 Department of Psychiatry, Institute of Biomedical Science, Tokushima University Graduate
School, Tokushima, Japan
6 Laboratory for Molecular Mechanisms of Brain Development, RIKEN Center for Brain
Science, Wako, Saitama, Japan
7 Agricultural Artificial Intelligence (AI) Research Office, Research Center for Agricultural
Information Technology, National Agriculture and Food Research Organization (NARO),
Tokyo, Japan
Running title: Cdh23 and prepulse inhibition
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted November 4, 2020. ; https://doi.org/10.1101/2020.10.29.360180doi: bioRxiv preprint

2
*Correspondence:
Takeo Yoshikawa, MD, PhD
Laboratory for Molecular Psychiatry
RIKEN Center for Brain Science
2-1 Hirosawa, Wako, Saitama 351-0198
Japan
Tel: +81(Japan)-48-467-5968
Fax: +81(Japan)-48-467-7462
E-mail: takeo.yoshikawa@riken.jp
Abstract: 187 words
Main text: 3876 words
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted November 4, 2020. ; https://doi.org/10.1101/2020.10.29.360180doi: bioRxiv preprint

3
Abstract
We previously identified quantitative trait loci (QTL) for prepulse inhibition (PPI), an
endophenotype of schizophrenia, on mouse chromosome 10 and reported Fabp7 as a
candidate gene from an analysis of F2 mice from inbred strains with high (C57BL/6N; B6)
and low (C3H/HeN; C3H) PPI levels. Here, we reanalyzed the previously reported QTLs
with increased marker density. The highest LOD score (26.66) peaked at a synonymous
coding and splice-site variant, c.753G>A (rs257098870), in the Cdh23 gene on chromosome
10; the c.753G (C3H) allele showed a PPI-lowering effect. Bayesian multiple QTL
mapping also supported the same variant with a posterior probability of 1. Thus, we
engineered the c.753G (C3H) allele into the B6 genetic background, which led to dampened
PPI. We also revealed an e-QTL (expression-QTL) effect imparted by the c.753G>A variant
for the Cdh23 expression in the brain. In a human study, a homologous variant (c.753G>A;
rs769896655) in CDH23 showed a nominally significant enrichment in individuals with
schizophrenia. We also identified multiple potentially deleterious CDH23 variants in
individuals with schizophrenia. Collectively, the present study reveals a PPI-regulating
Cdh23 variant and a possible contribution of CDH23 to schizophrenia susceptibility.
Keywords: prepulse inhibition, quantitative trait locus, Cdh23 (CDH23), schizophrenia,
hearing loss
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted November 4, 2020. ; https://doi.org/10.1101/2020.10.29.360180doi: bioRxiv preprint

4
Introduction
Deciphering neurobiological correlates for behavioral phenotypes in terms of genetic
predispositions that affect molecular, cellular and circuit-level functions is crucial for
understanding the pathogenesis of psychiatric disorders.
1
However, polygenicity and the
inherent limitations of psychiatric diagnosis based on the subjective experiences of patients
have impeded this endeavor.
2, 3
With efforts to classify psychiatric disorders based on the
dimensions of observable behavioral and neurobiological measures, endophenotype-based
approaches for elucidating genetic liability have attracted interest because these approaches
could mitigate clinical heterogeneity.
4, 5
Among the behavioral endophenotypes, the prepulse inhibition (PPI) of acoustic
startle response, a reflection of sensorimotor gating, has been consistently reported to be
dampened in psychiatric disorders, particularly in schizophrenia.
6, 7
As a robust and heritable
endophenotype in schizophrenia,
8-11
the genes/variants conferring the risk of dampened PPI
can help to elucidate the genetic architecture of schizophrenia.
12, 13
We have previously
performed a large-scale quantitative trait loci (QTL) analysis for PPI and mapped six major
loci through an analysis of 1,010 F2 mice derived by crossing selected inbred mouse strains
with high (C57BL/6NCrlCrlj: B6Nj) and low (C3H/HeNCrlCrlj: C3HNj) PPI.
14
Among these
six loci, the chromosome-10 QTL showed the highest logarithm of odds (LOD) score, and we
identified fatty acid-binding protein 7 (Fabp7) as one of the candidate genes in the PPI
regulation and schizophrenia pathogenesis.
14, 15
However, this high LOD score cannot be
explained by a single gene, considering the low phenotypic variance attributed to the
individual genes/markers, suggesting the presence of additional causative gene(s).
16
Therefore, in the current study, we aimed to (i) delineate additional gene(s) that regulate the
PPI phenotype by reanalyzing the QTL with higher marker density, (ii) validate the candidate
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted November 4, 2020. ; https://doi.org/10.1101/2020.10.29.360180doi: bioRxiv preprint

5
gene variant(s) by analyzing causal allele knock-in mice, and (iii) examine the potential role
of candidate gene in schizophrenia susceptibility.
Materials and Methods
Animals
All animal experiments were approved by the Animal Ethics Committee at RIKEN. For the
QTL analysis, we genotyped F2 mice from our previous study.
14
For other experiments,
C57BL/6NCrl (B6N) and C3H/HeNCrl (C3HN) mice were used (Japan's Charles River
Laboratories, Yokohama, Japan). Note that the previously used mouse strains
C57BL/6NCrlCrlj (B6Nj) and C3H/HeNCrlCrlj (C3HNj) are no longer available because
their supply ended in 2014.
16
Human DNA Samples
Studies involving human subjects were approved by the Human Ethics Committee at RIKEN.
All participants in the genetic studies gave informed written consent. For resequencing the all
protein-coding exons of the CDH23 gene, a total of 1,200 individuals with schizophrenia
(diagnosed according to DSM-IV) of Japanese descent (657 men, mean age 49.1±14.0 years;
543 women, mean age 51.1±14.4 years) were used. For subsequent genotyping, an additional
811 individuals with schizophrenia were included, for a total of 2,011 affected individuals
(1,111 men, mean age 47.2 ± 14.1 years; 901 women, mean age 49.2 ± 14.7 years), along
with 2,170 healthy controls (889 men, mean age 39.2 ±13.8 years; 1,281 women, mean age
44.6 ± 14.1 years). The samples were recruited from the Honshu region of Japan (the main
island), where the population fall into a single genetic cluster.
17, 18
Additionally, we also used
genetic data of 8,380 healthy Japanese controls from the Integrative Japanese Genome
Variation Database (iJGVD) by Tohoku Medical Megabank Organization (ToMMo)
.CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted November 4, 2020. ; https://doi.org/10.1101/2020.10.29.360180doi: bioRxiv preprint