Identification and functional studies of regulatory variants responsible for the association of NRG3 with a delusion phenotype in schizophrenia

Abstract

We previously reported genetic linkage for Schizophrenia (SZ) (NPL of 4.7) at 10q22 in the Ashkenazi Jewish (AJ) population. In follow up fine mapping we found strong evidence of association between three intronic single nucleotide variants (SNVs) in the 5' end of Neuregulin 3 (NRG3) and the delusion factor score of our phenotypic principal component analysis. Two independent groups replicated these findings, indicating that variants in NRG3 confer risk for a delusion-rich SZ subtype. To identify the causative variants, we sequenced the 162 kb linkage disequilibrium (LD) block covering the NRG3 5' end in 47 AJ SZ patients at the extremes of the delusion factor quantitative trait distribution. Among the identified variants we found 5 noncoding SNVs present on the high delusion factor haplotype and significantly overrepresented in high delusion factor subjects. We tested these for regulatory effects and found that risk alleles of rs10883866 and rs60827755 decreased and increased, respectively, the expression of a reporter gene as compared to the reference allele. In post-mortem brain RNA quantification experiments we found the same variants also perturb relative expression of alternative NRG3 isoforms. In summary, we have identified regulatory SNVs contributing to the association of NRG3 with delusion symptoms in SZ.

Keywords: Ashkenazi; NRG3; delusion; functional studies; neuregulin-3; psychiatric disease.

Fig. 1

Structure of NRG3. a Diagram of NRG3 exons as well as a proposed numerical nomenclature for validated exons in NRG3 (gray) located at 10q 23.1. Exons and introns are not to scale. b Table of the genomic coordinates on 10q (Start, Stop) and the size of each of the NRG3 exons. The column labeled ‘Kao’ correlates these exons with those described previously by Kao et al. [25]. * A shorter (3-bp) version of this exon has been observed [Vernon, unpubl. data]. ** A shorter (5-bp) version of this exon has been reported (EMBL:ABC69293.1.).

Fig. 2

a Diagram of NRG3 amplicons used for sequencing (top) and negative log₁₀ p values for differences in allele count between high and low delusion factor subjects (bottom). Variants with more alternative alleles in low delusion factor patients are in black and those with more alternative alleles in high delusion factor patients are in red. b Gel (0.75% agarose) electrophoresis of the NRG3 amplicons shown in a flanked by a 1-kb ladder size marker from a subject heterozygous for a microdeletion (hg19 chr10:83611664-83613507).

Fig. 3

Results from DLAs of the microdeletion region. Each construct was tested in 6 biological replicates per experiment and was normalized to the promoter-only vector. No statistically significant differences are observed between the DEL and the REF allele.

Fig. 4

DLAs of short ∼320-bp sequence segments surrounding the indicated SNPs. The constructs were expressed in HEK 293 cells (a), cultured rat primary cortical neurons (b), HT22 cells (c) and Neuro2A cells (d). Statistically significant results from two-tailed t tests are shown. Each SNP (or pair of SNPs, in the case of rs7919976 and rs10883862) is represented by a different color; the left bar in each set represents the REF allele and the right the risk allele. Each construct was tested in 6 biological replicates per experiment and normalized to the promoter-only vector.

Fig. 5

Effect of rs10748842 on NRG3 alternative transcripts. Box plots of levels of exon 1b transcripts relative to exon 1a transcripts for the different genotype groups and in the two brain regions examined. The homozygotes for the rare genotype CC were very few (2 and 1 for the STG and DLPFC, respectively) and grouped with the heterozygotes.