The impact of common variants on gene expression in the human brain: from RNA to protein to schizophrenia risk

Abstract

Background: The impact of genetic variants on gene expression has been intensely studied at the transcription level, yielding in valuable insights into the association between genes and the risk of complex disorders, such as schizophrenia (SCZ). However, the downstream impact of these variants and the molecular mechanisms connecting transcription variation to disease risk are not well understood.

Results: We quantitated ribosome occupancy in prefrontal cortex samples of the BrainGVEX cohort. Together with transcriptomics and proteomics data from the same cohort, we performed cis-Quantitative Trait Locus (QTL) mapping and identified 3,253 expression QTLs (eQTLs), 1,344 ribosome occupancy QTLs (rQTLs), and 657 protein QTLs (pQTLs) out of 7,458 genes quantitated in all three omics types from 185 samples. Of the eQTLs identified, only 34% have their effects propagated to the protein level. Further analysis on the effect size of prefrontal cortex eQTLs identified from an independent dataset showed clear post-transcriptional attenuation of eQTL effects. To investigate the biological relevance of the attenuated eQTLs, we identified 70 expression-specific QTLs (esQTLs), 51 ribosome-occupancy-specific QTLs (rsQTLs), and 107 protein-specific QTLs (psQTLs). Five of these omics-specific QTLs showed strong colocalization with SCZ GWAS signals, three of them are esQTLs. The limited number of GWAS colocalization discoveries from omics-specific QTLs and the apparent prevalence of eQTL attenuation prompted us to take a complementary approach to investigate the functional relevance of attenuated eQTLs. Using S-PrediXcan we identified 74 SCZ risk genes, 34% of which were novel, and 67% of these risk genes were replicated in a MR-Egger test. Notably, 52 out of 74 risk genes were identified using eQTL data and 70% of these SCZ-risk-gene-driving eQTLs show little to no evidence of driving corresponding variations at the protein level.

Conclusion: The effect of eQTLs on gene expression in the prefrontal cortex is commonly attenuated post-transcriptionally. Many of the attenuated eQTLs still correlate with SCZ GWAS signal. Further investigation is needed to elucidate a mechanistic link between attenuated eQTLs and SCZ disease risk.

Fig. 1.. Genetic regulation of gene expression in the human brain.

(A) P-value quantile-quantile plot between the observed (Y-axis) and the expected based on null distribution (X-axis). The black line indicates the expected distribution of p values when there are no real QTL signals. The number of cis-QTLs (i.e. the most significantly associated SNP for each gene) identified at 10% FDR is labeled in the top left inset. (B) Replication rate between QTL types. Proportions of QTLs replicated in the other two omics types are listed in the 3×3 matrix. Each row is a discovery omics type and each element of the row correspond to the proportion QTL signals replicated in the omics type specified by the column label. For example, only 34% of the eQTL signals were replicated in the protein data. (C) Effect size of CMC eQTL SNPs in BrainGVEX data. Mean and 95% confidence interval of absolute per allele effect across all CMC eQTL SNPs that were also analyzed in the BrainGVEX union set is shown.

Fig. 2.. Signal colocalization between schizophrenia GWAS and omics-specific QTLs.

(A, C) Boxplots summarizing normalized gene expression level stratified by QTL genotypes for CCDC117 esQTLs (A) and UGGT2 rsQTLs (C). (B, D) Manhattan plots showing p value distribution for each QTL type and schizophrenia GWAS for the 1Mb QTL mapping window flanking CCDC117 (B) and UGGT2 (D). The red line indicates the position of the lead colocalization SNP between omics-specific QTLs and schizophrenia GWAS.

Fig. 3.. Schizophrenia risk genes identified from each of the three omics types RNA-Seq (mRNA), ribo-seq (ribosome occupancy), and proteomics (protein) using S-PrediXcan.

(A) Manhattan plots showing significance level (i.e. -log10 FWER from S-PrediXcan) of gene-schizophrenia association across the genome for genes that pass the 5% FWER significance cutoff. The black horizontal dotted line indicated the significance cutoff. Risk genes are color-coded according to the MR test results. Grey asterisks mark the risk genes that failed the two-sample MR tests; dark red solid circle marks the risk genes that pass both one-sample MR tests (passing both one-sample MR tests suggest that transcriptionally regulated protein level differences between individuals drives the disease risk); blue triangle marks the risk genes that pass one of the two one-sample MR tests; orange rectangle marks the risk genes that failed both one-sample MR tests. (B) Venn diagram illustrating the number and corresponding percentage of overlapping risk genes between omics types. (C) Boxplots summarizing normalized gene expression level stratified by eQTL genotypes for KHK. (D) Manhattan plots showing p value distribution for each QTL type and schizophrenia GWAS for the 1Mb QTL mapping window flanking KHK. The red line indicates the position of the lead colocalization SNP between eQTLs and schizophrenia GWAS.

Fig. 4.. Signal colocalization between schizophrenia GWAS and eQTLs of example risk genes.

(A, C) Boxplots summarizing normalized gene expression level stratified by QTL genotypes for NEK4 eQTLs (A) and SF3B1 eQTLs (C). (B, D) Manhattan plots showing p value distribution for each QTL type and schizophrenia GWAS for the 1Mb QTL mapping window flanking NEK4 (B) and SF3B1 (D). The red line indicates the position of the colocalization lead SNP between eQTLs and schizophrenia GWAS.