Laser capture microdissection of human pancreatic islets reveals novel eQTLs associated with type 2 diabetes

Abstract

Objective: Genome wide association studies (GWAS) for type 2 diabetes (T2D) have identified genetic loci that often localise in non-coding regions of the genome, suggesting gene regulation effects. We combined genetic and transcriptomic analysis from human islets obtained from brain-dead organ donors or surgical patients to detect expression quantitative trait loci (eQTLs) and shed light into the regulatory mechanisms of these genes.

Methods: Pancreatic islets were isolated either by laser capture microdissection (LCM) from surgical specimens of 103 metabolically phenotyped pancreatectomized patients (PPP) or by collagenase digestion of pancreas from 100 brain-dead organ donors (OD). Genotyping (> 8.7 million single nucleotide polymorphisms) and expression (> 47,000 transcripts and splice variants) analyses were combined to generate cis-eQTLs.

Results: After applying genome-wide false discovery rate significance thresholds, we identified 1,173 and 1,021 eQTLs in samples of OD and PPP, respectively. Among the strongest eQTLs shared between OD and PPP were CHURC1 (OD p-value=1.71 × 10^-24; PPP p-value = 3.64 × 10^-24) and PSPH (OD p-value = 3.92 × 10^-26; PPP p-value = 3.64 × 10^-24). We identified eQTLs in linkage-disequilibrium with GWAS loci T2D and associated traits, including TTLL6, MLX and KIF9 loci, which do not implicate the nearest gene. We found in the PPP datasets 11 eQTL genes, which were differentially expressed in T2D and two genes (CYP4V2 and TSEN2) associated with HbA1c but none in the OD samples.

Conclusions: eQTL analysis of LCM islets from PPP led us to identify novel genes which had not been previously linked to islet biology and T2D. The understanding gained from eQTL approaches, especially using surgical samples of living patients, provides a more accurate 3-dimensional representation than those from genetic studies alone.

Keywords: Genetics; Islets; Laser capture microdissection; Type 2 diabetes; eQTLs.

Figure 1

Ancestry clustering and eQTL methodology. A. Genotype clustering of samples from PPP and OD samples confirmed European descent of subjects, compared to the 1,000 genomes. B. An overview of the methodology utilized to obtain eQTLs from islets of OD and PPP subjects were isolated by limited digestion from OD material and by LCM from PPP material. Acis-window of 500 kb was used with adjustment on gender and age and a false discovery rate (FDR) of <5% was used as a cut-off.

Figure 2

Overview of eQTL significant loci. A. Manhattan plot of the significant cis-eQTLs (≤500 kb), showing the top 10 significant eQTLs (FDR<5%), from 103 PPP and 100 OD subjects. B. Venn diagram summarising number of genes shared between PPP, OD, GWAS and previously identified eQTL genes in islets (Fadista et al., 2014; van de Bunt et al., 2015). C. An overview of the number of eQTL locations within putative regulatory regions within the genome that correspond with eQTL regions in OD and PPP datasets.

Figure 3

Biological interpretations of eQTL genes. Genes in our dataset that have been shown with IPA to have the MODY gene HNF4A as the upstream regulator for A. OD and B. PPP. Genes highlighted in purple have a relevant function in carbohydrate metabolism, lipid metabolism or diabetes.

Figure 4

Summary of eQTL genes with relevance to T2D. A schematic of a representative pancreatic beta cell, with identified genes annotated to their known sub-cellular localization using gene ontology. These genes include overlapping GWAS genes (black), eQTLs in LD with GWAS loci (asterisks), eQTL genes from the PPP dataset that were differentially expressed in T2D (red) or associated with HbA1c (pink), consistent in directional effect with eQTLs. (Figure was illustrated using app.biorender.io).