Long-read RNA sequencing reveals synaptic genes with differential isoform usage in schizophrenia [short-read RNAseq ]

Over the past 15+ years, genetic studies have been notably successful in revealing the architecture of multiple psychiatric disorders, including schizophrenia. It is now widely acknowledged that schizophrenia is highly polygenic, consisting of both common variants of unknown significance, identified through genome-wide association studies (GWAS), and rare CNVs and coding variants. Despite these advances, the underlying mechanisms of specific genes implicated by GWAS are not well understood. Multiple lines of evidence implicate alternative splicing in the pathophysiology of schizophrenia. Single-nucleotide polymorphisms (SNPs) within implicated schizophrenia loci could alter isoform diversity and abundances which may not be reflected in a typical differential expression study. Hence, we generated a comprehensive isoform survey of postmortem human dorsolateral prefrontal cortex (DLPFC) from schizophrenia cases and neurotypical controls to identify case-control isoform-level differences. We developed an analysis pipeline that combines the strengths of PacBio SMRT long-read RNA sequencing in conducting a detailed isoform census with the capacity of short-read RNA sequencing for the quantification of isoform abundances. From several hundred thousand discovered long-read isoforms we curated a transcriptome with tens of thousands of high confidence novel isoforms. We then identified differential isoform usage (DIU) genes using a combination of established and in-house pipelines that enables case-control comparisons. Many of these novel isoforms are differentially expressed in schizophrenia DLPFC vs neurotypical controls. Differentially expressed genes are enriched in gene sets related to synaptic structure and function, RNA binding and splicing, as well as cell types previously implicated in schizophrenia, including cortical excitatory neurons, medium spiny neurons, and pyramidal CA1 neurons. Publicly available splicing data, genotyping, proteomics, and single nucleus sequencing results verify and support our results.