1 Objective: Genome-wide association studies (GWAS) have successfully identified 145 loci 2 implicated in schizophrenia (SCZ). However, the underlying mechanisms remain largely 3 unknown. 4Methods: Here, we analyze 1,479 RNA-seq data from 13 postmortem brain regions in 5 combination with their genotype data and identify SNPs that are associated with expression 6 throughout the genome by dissecting expression features to genes (eGene) and exon-exon 7 junctions (eJunction). Then, we co-localize eGene and eJunction with SCZ GWAS using SMR and 8 mapping. Multiple ChIP-seq data and DNA methylation data generated from brain were used for 9 identifying the causative variants. Finally, we used a hypothesis-free (no SCZ risk loci considered) 10 enrichment analysis to determine implicated pathways. 11
Results:We identified 171 genes and eight splicing junctions located within four genes (SNX19, 12 ARL6IP4, APOPT1 and CYP2D6) that potentially contribute to SCZ susceptibility. Among the 13 genes, CYP2D6 is significantly associated with SCZ SNPs in both eGene and eJunction across the 14 13 brain regions. In-depth examination of the CYP2D6 region revealed that a non-synonymous 15 single nucleotide variant (SNV) rs16947 is strongly associated with a higher abundance of 16 CYP2D6 exon 3 skipping junctions. While we found rs133377 and two other SNPs in high linkage 17 disequilibrium (LD) with rs16947 (r 2 = 0.9539), histone acetylation analysis showed they are 18 located within active transcription start sites. Furthermore, our data-driven enrichment analysis 19 showed CYP2D6 is significantly involved in drug metabolism of tamoxifen, codeine and 20 citalopram. 21
Conclusions:Our study facilitates an understanding of the genetic architecture of SCZ and 22 provides new drug targets. 23