Transcriptomic Imputation of Bipolar Disorder and Bipolar subtypes reveals 29 novel associated genes

Abstract

60Bipolar disorder is a complex neuropsychiatric disorder presenting with episodic mood 61 disturbances. In this study we use a transcriptomic imputation approach to identify novel genes 62 and pathways associated with bipolar disorder, as well as three diagnostically and genetically 63 distinct subtypes. Transcriptomic imputation approaches leverage well-curated and publicly 64 available eQTL reference panels to create gene-expression prediction models, which may then 65 be applied to "impute" genetically regulated gene expression (GREX) in large GWAS datasets. 66By testing for association between phenotype and GREX, rather than genotype, we hope to 67 identify more biologically interpretable associations, and thus elucidate more of the genetic 68 architecture of bipolar disorder. 69 70We applied GREX prediction models for 13 brain regions (derived from CommonMind 71Consortium and GTEx eQTL reference panels) to 21,488 bipolar cases and 54,303 matched 72 controls, constituting the largest transcriptomic imputation study of bipolar disorder (BPD) to 73 date. Additionally, we analyzed three specific BPD subtypes, including 14,938 individuals with 74 subtype 1 (BD-I), 3,543 individuals with subtype 2 (BD-II), and 1,500 individuals with 75 schizoaffective subtype (SAB). 76 77We identified 125 gene-tissue associations with BPD, of which 53 represent independent 78 associations after FINEMAP analysis. 29/53 associations were novel; i.e., did not lie within 1Mb 79 of a locus identified in the recent PGC-BD GWAS. We identified 37 independent BD-I gene-80 tissue associations (10 novel), 2 BD-II associations, and 2 SAB associations. Our BPD, BD-I and 81 BD-II associations were significantly more likely to be differentially expressed in post-mortem 82 brain tissue of BPD, BD-I and BD-II cases than we might expect by chance. Together with our 83 pathway analysis, our results support long-standing hypotheses about bipolar disorder risk, 84 including a role for oxidative stress and mitochondrial dysfunction, the post-synaptic density, 85 and an enrichment of circadian rhythm and clock genes within our results. 86 87 88Bipolar disorder (BPD) is a serious episodic neuropsychiatric disorder presenting with extreme 89 elation, or mania, and severe depressive states 1 . In tandem, individuals with bipolar often 90 experience disturbances in thinking and behavior, as well as psychotic features such as 91 delusions and hallucinations 1 . Estimates of the prevalence of BPD within the general population 92 range from 0.5-1.5% 1,2 . Bipolar disorder is highly heritable, with siblings of probands at an 8-93 fold increased risk of the disorder 1,2 , and twin studies producing strikingly high estimates of 94 heritability, around 89-93% 1,3,4 . More recently, genetic studies of BPD have indicated SNP 95 heritability estimates of 17-23% 5 . 96 97

Deep transcriptome sequencing of subgenual anterior cingulate cortex reveals disorder-specific expression changes in major psychiatric disorders

“…Integration of our eGene data with the latest BD GWAS results detected 4 pleiotropic variants, increasing to 16 when all brain samples were considered ( Supplementary Table 13). This SMR analysis replicated 4 genes reported in previous SMR studies of BD 6,21,22 : DDHD2, MCHR1, FADS1, and PACS1. The present study detected 5 novel BD eGenes, including: ORMDL3, noted above in relation to SCZ; XPNPEP3, an ezyme involved in glomerular filtration, protein processing, and alternative splicing; PDF, a mitochondrial protein involved in cell proliferation; and 2 non-coding RNA genes.…”

“…For some genes, the same transcript was DE in all 3 disorders (e.g., NR4A1-213, Figure 4), but there were several instances of alternative isoform usage, where distinct transcripts of the same gene were DE in different disorders (e.g., HDAC11, NR4A1; Figure 4). [22][23][24]. Since the absolute difference between case groups is larger for opposing transcripts (Figure 5d-5e), when power is limited such transcripts should be more easily detected in case-case than in case-control comparisons.…”

Genome-wide association study of over 40,000 bipolar disorder cases provides new insights into the underlying biology

“…The top association in our GWAS was in the TRANK1 locus on chromosome 3, which has previously been implicated in BD 12,18,57 . Although BD-associated SNPs in this locus are known to regulate TRANK1 expression 74 , our eQTL analyses support a stronger but correlated regulation of DCLK3 , located 87 kb upstream of TRANK1 41,75 . Both FURIN and DCLK3 also encode druggable proteins (although they are not targets for any current psychiatric medications) 71,76 .…”

Identifying tissues implicated in Anorexia Nervosa using Transcriptomic Imputation

“…Using exclusively brain-gene association statistics as an input to our MAGMA analysis resulted in 51 significantly enriched pathways. 35/51 pathways were from the hypothesis-driven test; these included circadian entrainment (p=2.6×10 −04 ), addictive behaviors (nicotine, alcohol, cocaine, and morphine dependence, p<0.0045), calcium-gated voltage channels, and a large number of pathways related to processes in the post-synaptic density (Table 4), in line with pathway results from other psychiatric disorders 10,30,50,51 . A further 25 significantly enriched pathways were identified in the agnostic analysis, including further evidence of circadian entrainment (p=1.39×10 −06 ), long-term potentiation (p=4.44×10 −06 ), as well as multiple pathways implicating ear and neuronal system development in mice (p<1.2×10 −04 ).…”