Regulatory predictions for disease susceptible loci of bipolar disorder: a focus on non-coding RNAs

Prof. Po-Hsiu Kuo



Genome-wide association studies (GWAS) have found numerous loci to be associated with complex traits, including bipolar disorder (BD), a severe and highly heritable neuropsychiatric disorder. A substantial proportion of identified markers are mapped to regions with non-coding RNAs (ncRNAs). Recent studies have revealed the critical roles of ncRNAs on genomic regulation and neuronal and brain development. The present study addressed the potential involvement of ncRNAs in the etiology of BD through systematically empirical and in silico data integration. We also discussed the possible interaction mechanisms between ncRNAs and coding genes.



We first annotated BD-associated loci with Ensembl ncRNAs references, including micro-RNAs (miRNAs) and long non-coding RNAs (lncRNAs). We then investigated functional interpretations of mapped ncRNAs by integrating multiple databases to explore gene expression profiles in the specific brain regions and possible transcriptional regulation with transcription factors (TFs). Furthermore, weight gene co-expression network analysis was performed with “guilt by association” approach to discover the possible pathways that were involved with the candidate ncRNAs and previous coding-gene results. 



The proportion of BD-associated mapping loci increased from 64% to 94% with ncRNAs annotation. In total, 34 candidate ncRNAs were mapped. With brain tissue-specific expression patterns, we selected 7 mapped lncRNAs and 4 mapped miRNAs for further analysis. Three of the mapped miRNAs were predicted to interact with multiple lncRNAs and their involved pathways were highly related to psychiatric disorders. The TF-enrichment analysis revealed 6 TFs and their most common target was a candidate lncRNA (ENSG00000204929), which was ever reported for education attainment and cognitive ability. Co-expression network analysis showed that the candidate lncRNAs were related to “Oxidative phosphorylation”, “Mitochondrial inner membrane”, “Respiratory electron transport” and “MHC class II protein complex”. These were different from voltage-gated related pathways that were enriched by mapped coding genes. 



The present study systematically explored the possible involvement of ncRNAs from BD GWAS data and interpreted their possible transcriptional regulation ability. Our findings expand the current knowledge on the involvement of ncRNAs in BD, which represent a useful next step to provide insight into designing future experimental studies to explore the complex networks, pathogenesis and biological mechanisms underlying BD.