Genome-wide association study identifies two loci strongly affecting transferrin glycosylation

Abstract

Polysaccharide sidechains attached to proteins play important roles in cell-cell and receptor-ligand interactions. Variation in the carbohydrate component has been extensively studied for the iron transport protein transferrin, because serum levels of the transferrin isoforms asialotransferrin + disialotransferrin (carbohydrate-deficient transferrin, CDT) are used as biomarkers of excessive alcohol intake. We conducted a genome-wide association study to assess whether genetic factors affect CDT concentration in serum. CDT was measured in three population-based studies: one in Switzerland (CoLaus study, n = 5181) and two in Australia (n = 1509, n = 775). The first cohort was used as the discovery panel and the latter ones served as replication. Genome-wide single-nucleotide polymorphism (SNP) typing data were used to identify loci with significant associations with CDT as a percentage of total transferrin (CDT%). The top three SNPs in the discovery panel (rs2749097 near PGM1 on chromosome 1, and missense polymorphisms rs1049296, rs1799899 in TF on chromosome 3) were successfully replicated , yielding genome-wide significant combined association with CDT% (P = 1.9 × 10(-9), 4 × 10(-39), 5.5 × 10(-43), respectively) and explain 5.8% of the variation in CDT%. These allelic effects are postulated to be caused by variation in availability of glucose-1-phosphate as a precursor of the glycan (PGM1), and variation in transferrin (TF) structure.

Figure 1.

Local plots for the two regions associated with CDT%. The y-axis shows significance of the association as –log₁₀(P), x-axis is physical distance (on Build 36). Shading of the data points indicates the strength of LD with the most significant SNP in the region. (A) SNP associations with CDT (as a percentage of total transferrin) near the PGM1 locus on chromosome 1. The plotted combined P-values are from a meta-analysis of the discovery (CoLaus) and both (Australian) replication cohorts. (B) Association plot as described above, but for the region near the TF and SRPRB loci on chromosome 3. The plot shows combined P-values from a meta-analysis of the discovery (CoLaus) and the N-Latex-based (Australian) replication cohorts, showing univariate association results for all SNPs (red or uncoloured symbols). The three TF SNPs obtained by the multivariate model selection procedure (rs1049296, rs1799899 and rs8177318) are marked with green boundary circle.