Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 51 (6), 933-940

Multi-tissue Transcriptome Analyses Identify Genetic Mechanisms Underlying Neuropsychiatric Traits


Multi-tissue Transcriptome Analyses Identify Genetic Mechanisms Underlying Neuropsychiatric Traits

Eric R Gamazon et al. Nat Genet.


The genetic architecture of psychiatric disorders is characterized by a large number of small-effect variants1 located primarily in non-coding regions, suggesting that the underlying causal effects may influence disease risk by modulating gene expression2-4. We provide comprehensive analyses using transcriptome data from an unprecedented collection of tissues to gain pathophysiological insights into the role of the brain, neuroendocrine factors (adrenal gland) and gastrointestinal systems (colon) in psychiatric disorders. In each tissue, we perform PrediXcan analysis and identify trait-associated genes for schizophrenia (n associations = 499; n unique genes = 275), bipolar disorder (n associations = 17; n unique genes = 13), attention deficit hyperactivity disorder (n associations = 19; n unique genes = 12) and broad depression (n associations = 41; n unique genes = 31). Importantly, both PrediXcan and summary-data-based Mendelian randomization/heterogeneity in dependent instruments analyses suggest potentially causal genes in non-brain tissues, showing the utility of these tissues for mapping psychiatric disease genetic predisposition. Our analyses further highlight the importance of joint tissue approaches as 76% of the genes were detected only in difficult-to-acquire tissues.

Conflict of interest statement

Competing Interests Statement

The authors of this manuscript have no conflicts of interest to disclose.


Figure 1 |
Figure 1 |. PrediXcan and eQTL analysis of GWAS of psychiatric traits.
a, Heatmap and hierarchical clustering based on true positive rate (π1) for trait associations among PrediXcan associations in each tissue. Schizophrenia and bipolar disorder were most highly enriched in putamen basal ganglia and adrenal gland, respectively. Broad depression showed relatively strong enrichment in colon. b, Significant replication of schizophrenia PrediXcan associations (Bonferroni-adjusted P < 0.05, indicated by horizontal line) was observed within the UK Biobank sample (1,561 cases; 267,494 controls). Replication P-value for a gene was from the application of S-PrediXcan to GWAS summary statistics derived from BOLT-LMM. c, Heatmap and hierarchical clustering based on true positive rate (π1) for trait associations among b-eQTL associations in each tissue. d, Non-brain eQTL enrichment was observed even after conditioning on brain eQTLs from the joint-tissue eQTL analysis. The joint-tissue analysis allowed us to quantify the tissue specificity and tissue-sharedness of an eQTL-gene pair while taking into account differential power for eQTL discovery between tissues. The x-axis shows the π1 after including only those b-eQTLs active in the non-brain tissue but not regulatory in any of the 10 brain regions. The y-axis shows the π1 without the filtering. Points below the diagonal line show disease and tissue pairs for which a higher true positive rate was observed after filtering the brain eQTLs. Sample sizes: schizophrenia (40,675 cases, 64,643 controls); ADHD (20,183 cases, 35,191 controls); bipolar disorder (11,974 cases, 51,792 controls); broad depression (113,769 cases, 208,811 controls).
Figure 2 |
Figure 2 |. Proposing causal variant and causal gene mechanism at known schizophrenia-associated loci (n = 145 loci; 179 index SNPs).
a, Distribution of number of nearby genes (defined as +/− 1 Mb of index SNPs) at known loci. b, From the joint-tissue (METASOFT) analysis, 290 genes at known loci were significantly associated with schizophrenia (40,675 cases, 64,643 controls) in a tissue-dependent manner. Significant associations (METASOFT m-value ≥ 0.90) of these 290 genes are shown as red bars, while non-significant associations are shown as white bars. The figure shows clustering of genes that are shared across tissues, but also indicates genes that have tissue-specific effects. c, LocusZoom plot of the region surrounding the C4A gene indicates that C4A is significantly associated in GTEx cerebellum but not in dorsolateral prefrontal cortex. y-axis is the schizophrenia association P-value (in log scale). The large number of SNPs and genes in the locus illustrate the challenges in fine-mapping the causal variant and in determining the gene driver(s).
Figure 3 |
Figure 3 |. Complexity of identifying the relevant gene mechanism in tissue of pathology.
a, There is extensive eQTL sharing among tissues, as estimated by π1^. Note that the easily accessible tissue whole blood is a clear outlier, showing the least amount of sharing with the other tissues. Among the brain regions, cerebellum and cerebellar hemisphere are outliers. Sample sizes for the tissues can be found in Supplementary Table 1. b, Illustration of tissue-specific regulation from the METASOFT analysis. The b-eQTL of GATA2DA (rs2905432) impacts its expression only in whole blood, while the b-eQTL of PBX4 (chr19_19756073_D) influences its expression in cerebellum/cerebellar hemisphere and in hippocampus, but not in the remaining tissues. These examples indicate that delineating the biological mechanisms of schizophrenia based on eQTL information requires a tissue-specific approach for at least some of the eQTL-gene associations.
Figure 4 |
Figure 4 |. Genetically determined co-expression networks of disease-associated GReX.
a, We observed significant correlations in GReX of disease-associated genes (Bonferroni-adjusted P <0.05 from S-PrediXcan analysis of schizophrenia) for all tissues. Shown here are the pairwise GReX correlations in putamen basal ganglia, which had the largest π1^ for schizophrenia (40,675 cases, 64,643 controls). b, The observed GReX correlations were significantly greater (empirical P < 0.001) than expected based on 1,000 randomly generated, genetically defined co-expression networks. c, Hierarchical clustering of GReX of disease-associated genes in putamen basal ganglia was performed and identified several genetically determined clusters that appear to have coordinated expression including on distinct chromosomes.
Figure 5 |
Figure 5 |. Tissue-specific GReX, joint-tissue eQTL mapping, and tissue-specific or tissue-shared functional categories for schizophrenia.
a, Substantial concordance between tissues (for example, between whole blood (n = 338) and putamen basal ganglia (n = 82), Spearman ρ = 0.92, P = 3.86 × 10−164) was observed in the proportion of disease-associated GReX levels (Bonferroni-adjusted P < 0.05 from S-PrediXcan analysis) that map to a biological process (as defined by Gene Ontology). b, List of biological processes present only in whole blood and represented by five or more disease-associated GReX levels in the tissue. c, Enriched biological processes (hypergeometric test, Bonferroni-adjusted P < 0.05) in putamen basal ganglia (n = 82), which had the largest π1^ for schizophrenia, and the gene ratio for the genes that map to the processes.

Similar articles

See all similar articles

Cited by 3 articles


    1. Gratten J, Wray NR, Keller MC & Visscher PM Large-scale genomics unveils the genetic architecture of psychiatric disorders. Nature neuroscience 17, 782–790, doi:10.1038/nn.3708 (2014). - DOI - PMC - PubMed
    1. Nicolae DL et al. Trait-associated SNPs are more likely to be eQTLs: annotation to enhance discovery from GWAS. PLoS genetics 6, e1000888, doi:10.1371/journal.pgen.1000888 (2010). - DOI - PMC - PubMed
    1. Schizophrenia Working Group of the Psychiatric Genomics Consortium. Biological insights from 108 schizophrenia-associated genetic loci. Nature 511, 421–427, doi:10.1038/nature13595 (2014). - DOI - PMC - PubMed
    1. Gamazon ER et al. The convergence of eQTL mapping, heritability estimation and polygenic modeling: emerging spectrum of risk variation in bipolar disorder. arxiv (2013).
    1. Geschwind DH & Flint J Genetics and genomics of psychiatric disease. Science 349, 1489–1494, doi:10.1126/science.aaa8954 (2015). - DOI - PMC - PubMed

Methods-only references

    1. International Consortium for Blood Pressure Genome-Wide Association Studies et al. Genetic variants in novel pathways influence blood pressure and cardiovascular disease risk. Nature 478, 103–109, doi:10.1038/nature10405 (2011). - DOI - PMC - PubMed
    1. Shi H, Kichaev G & Pasaniuc B Contrasting the Genetic Architecture of 30 Complex Traits from Summary Association Data. American journal of human genetics 99, 139–153, doi:10.1016/j.ajhg.2016.05.013 (2016). - DOI - PMC - PubMed
    1. Gamazon ER, Cox NJ & Davis LK Structural architecture of SNP effects on complex traits. American journal of human genetics 95, 477–489, doi:10.1016/j.ajhg.2014.09.009 (2014). - DOI - PMC - PubMed
    1. Barbeira A et al. Integrating tissue specific mechanisms into GWAS summary results. Biorxiv, doi:10.1101/045260 (2017). - DOI
    1. Loh PR et al. Efficient Bayesian mixed-model analysis increases association power in large cohorts. Nature genetics 47, 284–290, doi:10.1038/ng.3190 (2015). - DOI - PMC - PubMed

Publication types