The splice site variant rs11078928 may be associated with a genotype-dependent alteration in expression of GSDMB transcripts

Abstract

Background: Many genetic variants have been associated with susceptibility to complex traits by genome wide association studies (GWAS), but for most, causal genes and mechanisms of action have yet to be elucidated. Using bioinformatics, we identified index and proxy variants associated with autoimmune disease susceptibility, with the potential to affect splicing of candidate genes. PCR and sequence analysis of whole blood RNA samples from population controls was then carried out for the 8 most promising variants to determine the effect of genetic variation on splicing of target genes.

Results: We identified 31 splice site SNPs with the potential to affect splicing, and prioritised 8 to determine the effect of genotype on candidate gene splicing. We identified that variants rs11078928 and rs2014886 were associated with altered splicing of the GSDMB and TSFM genes respectively. rs11078928, present in the asthma and autoimmune disease susceptibility locus on chromosome 17q12-21, was associated with the production of a novel Δ exon5-8 transcript of the GSDMB gene, and a separate decrease in the percentage of transcripts with inclusion of exon 6, whereas the multiple sclerosis susceptibility variant rs2014886, was associated with an alternative TFSM transcript encompassing a short cryptic exon within intron 2.

Conclusions: Our findings demonstrate the utility of a bioinformatic approach in identification and prioritisation of genetic variants effecting splicing of their host genes, and suggest that rs11078928 and rs2014886 may affect the splicing of the GSDMB and TSFM genes respectively.

Figure 1

Bioinformatic pipeline used to predict splice site SNPs that are associated with autoimmune diseases and inflammatory traits. Index inflammatory SNPs were identified that had been associated by GWAS with susceptibility to autoimmune diseases and inflammatory traits. The proxies to these GWAS SNPs were pulled using the SNAP Proxy Search tool (Broad Institute). The ‘functional consequence to transcript’ for each SNP was identified as being ‘Splice Site’ or ‘Essential Splice Site’ using the Biomart function of Ensembl. These SNPs were then bioinformatically analysed to predict whether a splicing change was likely to occur, using the programmes ESE finder (web-based tool that predicts ESE element sequences that are bound by SR proteins), NNSplice (web-based tool that algorithmically predicts core spice site sequences in a given sequence) and Alamut (splicing mutation prediction programme that amalgamates predictions from five different splice site prediction algorithms to identify potential core splice sites in a given gene transcript). A subset of SNPs was then prioritised for further analysis of their splicing by RT-PCR. The numbers in brackets denote how many SNPs were identified at each stage.

Figure 2

Enrichment of variants associated with inflammatory or autoimmune phenotypes in splicing control regions. We assessed the number of randomly selected variants located in splicing control elements by chance. For each set of 338 ‘index’ SNPs and their proxies, the percentage of variants located in ‘splice region elements’ is given on the X-axis (simulated percentage), and the number of SNPs at each percentage (count) is given on the Y-axis. The observed number of autoimmune or inflammatory SNPs located in ‘splice region elements’ is given by the line.

Figure 3

Wild-type (WT) and Novel Variant (NV) isoforms of the GSDMB gene and their expression with genotype. A. Showing exons 4–9 of the four RefSeq isoforms and the novel GSDMB transcipt (GSDMB NV), which is missing exons 5–8. Half the Reference Sequence (WT) transcripts include exon 6 and two are lacking exon 6; NM001165958.1 is the full length transcript. The star shows the position of rs11078928 (acceptor splice site of intron 5). The location of the Custom Taqman Assay Primers (indicated by an arrow) and probe (indicated by a rectangle) are shown. B. Chart showing the expression of the novel transcript GSDMB NV by genotype. Homozygotes for the minor allele show negligible expression of the novel transcript. Expression is normalised to the endogenous control RPLPO, and is shown relative to the expression of GSDMB NV in heterozygotes. Significant results (P < 0.05) are indicated by an asterix. C. Chart showing the expression with genotype of the GSDMB transcripts which include exon 6 (NM001165958.1, NM001165959.1). Homozygotes for the minor allele show no expression of exon 6-containing transcripts. Expression is normalised to the endogenous control RPLPO, and is shown relative to the expression of exon 6 GSDMB transcripts in heterozygotes. D-E. Charts showing total expression of GSDMB (all RefSeq isoforms, including novel) and expression of GSDMB WT (all RefSeq isoforms). Expression is normalized to the endogenous control RPLPO, and is shown relative to total and WT expression respectively in heterozygotes. Both show a decrease in expression with the minor allele, although these results did not reach statistical significance.

Figure 4

WT and NV isoforms of TSFM and their expression with genotype. A. Showing exons 2–3 of the RefSeq transcript (there are four RefSeq isoforms for TSFM, however, they are identical over this region of the gene) and the novel TSFM transcript (TSFM NV), which has an alternate 38 bp exon included between introns 2 and 3. The star shows the relative position of rs2014886 (introducing an intronic donor splice site within intron 2. B. Chart showing the expression of the novel transcript TSFM NV by genotype. Homozygotes for the major allele show negligible expression of the novel transcript. Expression is normalised to the endogenous control RPLPO, and is shown relative to the expression of TSFM NV in heterozygotes. Statistical significance is indicated by an asterix.

Figure 5

Showing the secondary structure change with the major A allele (right) and the minor allele G (left), as predicted by Mfold (http://mfold.rna.albany.edu/?q=mfold). The arrows indicate the position of the variant rs11078928 in the GSDMB transcript sequence.