Frontiers in Genetics (Apr 2021)

Meta-Analyses of Splicing and Expression Quantitative Trait Loci Identified Susceptibility Genes of Glioma

  • C. Pawan K. Patro,
  • Darryl Nousome,
  • The Glioma International Case Control Study (GICC),
  • Rose K. Lai,
  • Elizabeth B. Claus,
  • Dora Il’yasova,
  • Joellen Schildkraut,
  • Jill S. Barnholtz-Sloan,
  • Sara H. Olson,
  • Jonine L. Bernstein,
  • Christoffer Johansen,
  • Robert B. Jenkins,
  • Beatrice S. Melin,
  • Margaret R. Wrensch,
  • Richard S. Houlston,
  • Melissa L. Bondy

DOI
https://doi.org/10.3389/fgene.2021.609657
Journal volume & issue
Vol. 12

Abstract

Read online

BackgroundThe functions of most glioma risk alleles are unknown. Very few studies had evaluated expression quantitative trait loci (eQTL), and insights of susceptibility genes were limited due to scarcity of available brain tissues. Moreover, no prior study had examined the effect of glioma risk alleles on alternative RNA splicing.ObjectiveThis study explored splicing quantitative trait loci (sQTL) as molecular QTL and improved the power of QTL mapping through meta-analyses of both cis eQTL and sQTL.MethodsWe first evaluated eQTLs and sQTLs of the CommonMind Consortium (CMC) and Genotype-Tissue Expression Project (GTEx) using genotyping, or whole-genome sequencing and RNA-seq data. Alternative splicing events were characterized using an annotation-free method that detected intron excision events. Then, we conducted meta-analyses by pooling the eQTL and sQTL results of CMC and GTEx using the inverse variance-weighted model. Afterward, we integrated QTL meta-analysis results (Q < 0.05) with the Glioma International Case Control Study (GICC) GWAS meta-analysis (case:12,496, control:18,190), using a summary statistics-based mendelian randomization (SMR) method.ResultsBetween CMC and GTEx, we combined the QTL data of 354 unique individuals of European ancestry. SMR analyses revealed 15 eQTLs in 11 loci and 32 sQTLs in 9 loci relevant to glioma risk. Two loci only harbored sQTLs (1q44 and 16p13.3). In seven loci, both eQTL and sQTL coexisted (2q33.3, 7p11.2, 11q23.3 15q24.2, 16p12.1, 20q13.33, and 22q13.1), but the target genes were different for five of these seven loci. Three eQTL loci (9p21.3, 20q13.33, and 22q13.1) and 4 sQTL loci (11q23.3, 16p13.3, 16q12.1, and 20q13.33) harbored multiple target genes. Eight target genes of sQTLs (C2orf80, SEC61G, TMEM25, PHLDB1, RP11-161M6.2, HEATR3, RTEL1-TNFRSF6B, and LIME1) had multiple alternatively spliced transcripts.ConclusionOur study revealed that the regulation of transcriptome by glioma risk alleles is complex, with the potential for eQTL and sQTL jointly affecting gliomagenesis in risk loci. QTLs of many loci involved multiple target genes, some of which were specific to alternative splicing. Therefore, quantitative trait loci that evaluate only total gene expression will miss many important target genes.

Keywords