GLRB variants regulate nearby gene expression in human brain tissues

A recent genome-wide association study (GWAS) identified four genetic variants rs78726293, rs191260602, rs17035816 and rs7688285 in GLRB gene to be associated with panic disorder (PD) risk. In fact, GWAS is an important first step to investigate the genetics of human complex diseases. In order to translate into opportunities for new diagnostics and therapies, we must identify the genes perturbed by these four variants, and understand how these variant functionally contributes to the underlying disease pathogenesis. Here, we investigated the effect of these four genetic variants and the expression of three nearby genes including PDGFC, GLRB and GRIA2 in human brain tissues using the GTEx (version 6) and Braineac eQTLs datasets. In GTEx (version 6) dataset, the results showed that both rs17035816 and rs7688285 variants could significantly regulate PDGFC and GLRB gene expression. In Braineac dataset, the results showed that rs17035816 variant could significantly regulate GLRB and GRIA2 gene expression. We believe that these findings further provide important supplementary information about the regulating mechanisms of rs17035816 and rs7688285 variants in PD risk.


Results
eQTLs analysis in GTEx dataset. In GTEx (version 6) dataset, we found that rs78726293 and rs191260602 were not available in the. We then focused on the two genetic variants including rs17035816 and rs7688285 and three genes including PDGFC, GLRB and GRIA2 in the following analysis. The results indicated that both rs17035816 and rs7688285 variants could significantly regulate nearby gene expression in human brain tissues (significance threshold 0.05) 10 . In brief, rs17035816 variant could significantly regulate PDGFC gene expression in cerebellar hemisphere tissue (P = 8.70E-03), putamen basal ganglia tissue (P = 4.14E-03) and cerebellum tissue (P = 3.17E-02), and regulate GLRB gene expression in cerebellar hemisphere tissue (P = 1.50E-03). The rs7688285 variant could significantly regulate PDGFC gene expression in hippocampus tissue (P = 2.03E-02) and putamen basal ganglia tissue (P = 1.37E-02), and regulate GLRB gene expression in hypothalamus tissue (P = 2.63E-02). We further performed a multiple testing correction using a FDR threshold of 0.05 in these 10 brain tissues. Interestingly, rs17035816 variant still significantly regulates nearby gene expression after the multiple hypothesis test correction. More detailed results are described in Table 1. eQTLs analysis in Braineac dataset. In Braineac dataset, we found that rs78726293, rs191260602 and rs7688285 were not available. We then focused on the rs17035816 variant and three genes including PDGFC, GLRB and GRIA2 in the following analysis. The results showed that rs17035816 variant could significantly regulate nearby gene expression in human brain tissues (significance threshold 0.05). In brief, rs17035816 variant could significantly regulate GLRB and GRIA2 gene expression in cerebellar cortex tissue with P = 1.49E-02 and P = 3.49E-02, respectively. More detailed results are described in Table 2.
Meta-analysis. We further performed a meta-analysis in the same brain tissues including Cerebellum, Hippocampus, Frontal cortex, and Putamen. The results showed that rs17035816 variant could significantly regulate GLRB gene expression in cerebellum tissue. More detailed results are described in Table 3.

Discussion
Deckert et al. highlighted four genetic variants rs78726293, rs191260602, rs17035816 and rs7688285 in GLRB gene to be associated with PD risk 1 . In fact, GWAS is an important first step to investigate the genetics of human complex diseases as widely described in previous studies [11][12][13][14][15][16][17][18][19][20][21][22][23][24][25] . In order to translate into opportunities for new diagnostics and therapies, we must identify the genes perturbed by these four variants, and understand how these variant functionally contributes to the underlying disease pathogenesis [3][4][5][6][7][8]12,[26][27][28][29][30] . If a genetic variant is associated with increased or decreased expression of a particular gene, this suggests that the gene on which the variant acts could be in the causal pathway 31 . However, Deckert et al. revealed no significant cis-eQTL using the online GTEx database 1 . Here, we successfully identified significant cis-eQTL using all the original SNP-gene summary association results in the GTEx (version 6), even after the multiple hypothesis test correction using FDR threshold of 0.05.
In the GTEx dataset, we confirmed previous findings. Deckert et al. analyzed the post-mortem brain samples of 76 individuals, and identified that the rs7688285 A allele could significantly regulate increased mean expression of GLRB with beta = 0.498 and P = 0.013 1 . Here, our findings showed that rs7688285 variant A allele could significantly regulate increased PDGFC gene expression in hippocampus tissue (beta = 0.296 and P = 2.03E-02), reduced PDGFC gene expression in putamen basal ganglia tissue (beta = −0.297 and P = 1.37E-02), and increased GLRB gene expression in hypothalamus tissue (beta = 0.192 and P = 2.63E-02). Meanwhile, the results also showed some novel findings. Take rs17035816 variant for example, it could significantly regulate nearby gene expression even after the multiple hypothesis test correction as described in Table 1.
We further evaluated the potential association between these four genetic variants and the expression of three nearby genes including PDGFC, GLRB and GRIA2 in the Braineac dataset including 10 brain regions from 134 neuropathologically normal individuals of European descent 32 . Interestingly, the rs17035816 could significantly regulate increased GLRB and GRIA2 gene expression in cerebellar cortex. We believe that these findings further provide important supplementary information about the regulating mechanisms of rs17035816 and rs7688285 variants in PD risk in the human brain tissues. Genetic variants may need tissue, cell, region, disease specific factors to exert their influences on gene expression 33,34 . Here, we identified different results in GTEx (version 6) and Braineac eQTLs datasets. We think that disease status may influence the association between these genetic variants and GLRB and GRIA2 gene expression.
Here, the two variants rs78726293 and rs191260602 are not available in the GTEx dataset. Three variants rs78726293, rs191260602, and rs7688285 are not available in the Braineac dataset. We have used HaploReg (version 4) to identify the proxy SNPs based on the linkage disequilibrium (LD) information in 1000 Genomes Project (EUR) with r 2 ≥ 0.8 2 . However all these tagged SNPs are still not available in the GTEx dataset and the Braineac dataset. We think that following studies should further evaluate these potential expression associations using other eQTLs datasets in human brain regions.

Materials and Methods
The GTEx dataset. The GTEx (version 6) eQTLs dataset included 53 tissues, 544 donors and 8555 samples 35 . These 544 donors have several death pathologies including traumatic injury, cerebrovascular disease, heart disease, liver, renal, respiratory, and neurological diseases 35 . Here, we selected 10 human brain tissues including anterior cingulate cortex, caudate basal ganglia, cerebellar hemisphere, cerebellum, cortex, frontal cortex BA9, hippocampus, hypothalamus, nucleus accumbens basal ganglia, and putamen basal ganglia, which include at least 70 samples 10 . The GTEx used the RNA-Seq method to measure the gene expression 10 .
The Braineac dataset. The Braineac eQTLs dataset is from a web server for data from the UK Brain Expression Consortium (UKBEC) 32 . This dataset includes 10 brain regions from 134 neuropathologically normal individuals of European descent 32 . The 10 brain regions are cerebellar cortex, frontal cortex, hippocampus, medulla, occipital cortex, putamen, substantia nigra, temporal cortex, thalamus, and intralobular white matter 32 . The Braineac used the Affymetrix GeneChip Human exon 1.0 ST arrays to measure the gene expression 32 . The gene expression in transcript level is the Winsorised mean over exon-specific levels 32 .   Table 3. Meta-analysis of GTEx and Braineac datasets in four brain tissues. rs17035816, chr4:158088464, A/G; Significance level for a potential association is 0.05; Beta is the regression coefficient based on the effect allele. Beta > 0 and Beta < 0 means that this effect allele regulates increased and reduced gene expression, respectively. SE, standard error.
including genotyping PCs, genotyping array platform, 15, 30 or 35 PEER factors, and the gender 10 . Here, we downloaded all the SNP-gene association summary results from the GTEx (version 6) database to directly evaluate the potential association between these four genetic variants and gene expression of nearby genes. In the Braineac dataset, a linear regression analysis was also applied to evaluate the potential association between genetic variants and the expression of nearby genes using the R package Matrix EQTL 32 . Here, we downloaded the gene expression data and the genotype data of genetic variants with 1 Mb upstream of transcription start site and 1 Mb downstream of transcription end site from the Braineac online database 32 . We then evaluated the potential SNP-gene expression association using the R program. More detailed information is described in a recent study 36 .

Meta-analysis.
In the same brain tissue, we performed a meta-analysis to increase the statistical power.