Long Non-Coding RNA Dysregulation in Autism Spectrum Disorders (ASD)

Long non-coding RNAs (lncRNAs) can function as key regulators of gene expression, yet their expression and splicing patterns have not been fully explored in tissues relevant to ASD and other neuropsychiatric disorders. Here we used single-molecule real-time isoform sequencing of total RNA (IsoSeq) and lncRNA-capture enriched samples (Capture-IsoSeq) to construct a comprehensive map of full-length transcript isoforms in 10 human postmortem prefrontal cortex tissues. Transcript isoforms were classified based on their coding potential and conservation, and further characterized by integrating with 5’-end Seq and epigenetic profiles. We discovered >300 novel brain expressed lncRNA loci, as well as ~9,000 and ~8,000 novel high-confidence multiexonic lncRNA and mRNA isoforms, respectively. We further merged our IsoSeq maps with transcript models derived from RNA-Seq assemblies of 2,000 PsychENCODE cortical tissue samples, resulting in the most comprehensive map of both coding and non-coding transcripts in the cortex to date. Altogether, our map significantly increases the number of isoforms of brain-expressed genes, expands most of the lncRNA genes detected in our dataset comprising both 3’ and 5’ extensions, and reveals multiple cases of interleaved lncRNA-mRNA transcripts. Our augmented reference map was further used to quantify lncRNA expression by capture RNA-Seq (Capture-Seq) of 140 postmortem samples of prefrontal cortex and cerebellum from 35 ASD and 36 control subjects. Capture-Seq increased the lncRNA fraction of our RNA-Seq datasets from 5% to 57%, substantially improving our ability to assess changes in this transcript population. We identified multiple differentially expressed lncRNAs in ASD compared to controls, including neighbors of cis-regulated ASD-risk genes. Finally, coexpression network and “guilt-by-association” analyses further revealed genes under putative regulation of lncRNAs that are in pathways dysregulated in ASD. Taken together, our data constitute a valuable resource for integration with genetic risk variants and genomic data for discovery of candidate risk genes in ASD and other neuropsychiatric disorders.