RNA-sequencing of the brain transcriptome implicates dysregulation of neuroplasticity, circadian rhythms and GTPase binding in bipolar disorder

Abstract

RNA-sequencing (RNA-seq) is a powerful technique to investigate the complexity of gene expression in the human brain. We used RNA-seq to survey the brain transcriptome in high-quality postmortem dorsolateral prefrontal cortex from 11 individuals diagnosed with bipolar disorder (BD) and from 11 age- and gender-matched controls. Deep sequencing was performed, with over 350 million reads per specimen. At a false discovery rate of <5%, we detected five differentially expressed (DE) genes and 12 DE transcripts, most of which have not been previously implicated in BD. Among these, Prominin 1/CD133 and ATP-binding cassette-sub-family G-member2 (ABCG2) have important roles in neuroplasticity. We also show for the first time differential expression of long noncoding RNAs (lncRNAs) in BD. DE transcripts include those of serine/arginine-rich splicing factor 5 (SRSF5) and regulatory factor X4 (RFX4), which along with lncRNAs have a role in mammalian circadian rhythms. The DE genes were significantly enriched for several Gene Ontology categories. Of these, genes involved with GTPase binding were also enriched for BD-associated SNPs from previous genome-wide association studies, suggesting that differential expression of these genes is not simply a consequence of BD or its treatment. Many of these findings were replicated by microarray in an independent sample of 60 cases and controls. These results highlight common pathways for inherited and non-inherited influences on disease risk that may constitute good targets for novel therapies.

Figure 1

Significant correlation of differential gene expression in two independent sets of dorsolateral prefrontal cortex (DLPFC) brain samples. Fold change values are shown on the log₂ scale for both National Institutes of Health (NIH) Intramural Sequencing Center (NISC)-1 (x axis) and NISC2 (y axis). Each point represents one gene. Colored lines represent 5% density contours. The blue diagonal line represents the line of identity.

Figure 2

A circos plot showing diferentially expressed (DE) genes and transcripts in RNA-sequencing (RNA-seq) and microarray data, direction of expression, Gene Ontology (GO) term categories and chromosomal location. Each circle from the periphery to the core represents the following: chromosomal location; in yellow background: 1225 DE genes in RNA-seq; in gray background: 332 DE genes in microarray that overlap with RNA-seq (gene and/or transcript level); in yellow background: 2776 DE transcripts from RNA-seq, with red indicating upregulation and blue indicating downregulation at nominal P<0.05, and the height of the bars represents relative expression levels; in gray background: genome-wide association studies (GWAS) single-nucleotide polymorphisms (SNPs) linked to DE genes and transcripts; at the core, green radii connect genes involved in ion binding and black radii connect genes involved in regulation of cell development.

Figure 3

(a, b) Volcano plots generated by meta-analysis of differential gene expression at the level of gene and transcript. The x axis shows the weighted average meta log₂ fold change across both National Institutes of Health (NIH) Intramural Sequencing Center (NISC)-1 and NISC2 samples. The meta-analysis P-value (-log base 10) for differential expression is plotted on the y axis. At a nominal meta-analysis P-value< 0.05, downregulated genes (blue dots) and upregulated genes (red dots) are represented. The larger dots indicate genes DE at false discovery rate (FDR) <5% (corresponding to a P-value <0.000012). All other genes that passed quality control are shown as gray dots.