Associations Between Common and Rare Exonic Genetic Variants and Serum Levels of 20 Cardiovascular-Related Proteins: The Tromsø Study

Abstract

Background: Genetic variation can be used to study causal relationships between biomarkers and diseases. Here, we identify new common and rare genetic variants associated with cardiovascular-related protein levels (protein quantitative trait loci [pQTLs]). We functionally annotate these pQTLs, predict and experimentally confirm a novel molecular interaction, and determine which pQTLs are associated with diseases and physiological phenotypes.

Methods and results: As part of a larger case-control study of venous thromboembolism, serum levels of 51 proteins implicated in cardiovascular diseases were measured in 330 individuals from the Tromsø Study. Exonic genetic variation near each protein's respective gene (cis) was identified using sequencing and arrays. Using single site and gene-based tests, we identified 27 genetic associations between pQTLs and the serum levels of 20 proteins: 14 associated with common variation in cis, of which 6 are novel (ie, not previously reported); 7 associations with rare variants in cis, of which 4 are novel; and 6 associations in trans. Of the 20 proteins, 15 were associated with single sites and 7 with rare variants. cis-pQTLs for kallikrein and F12 also show trans associations for proteins (uPAR, kininogen) known to be cleaved by kallikrein and with NTproBNP. We experimentally demonstrate that kallikrein can cleave proBNP (NTproBNP precursor) in vitro. Nine of the pQTLs have previously identified associations with 17 disease and physiological phenotypes.

Conclusions: We have identified cis and trans genetic variation associated with the serum levels of 20 proteins and utilized these pQTLs to study molecular mechanisms underlying disease and physiological phenotypes.

Keywords: biomarker; coronary artery disease; exome; human; protein; venous thromboembolism.

Figure 1.

Overview of the 3 stages of association analyses. A, cis: for each of the 51 phenotypes (protein levels), we tested the variants located in the cis gene loci for associations with their respective protein level, (B) cis-acting-in-trans: we tested the significant cis-protein quantitative trait loci (pQTLs) from stage 1 for trans effects against each of the 50 other protein levels, and (C) trans: we tested all variants in the 50 cis loci (C3 and C3b share the same locus) for association with each of the 51 protein levels.

Figure 2.

Association of cis variants with protein levels. Modified Manhattan plot showing the –log10 P values for association between variants in each cis locus (interval encoding protein ±500 kb) and the respective protein levels. The red dashed line indicates the study-wide significant P-value cutoff when only examining cis regions (6.9×10⁻⁷) for a family-wise error rate <0.05.

Figure 3.

Schematic diagram showing proteins with identified trans associations and their nominal associations with variants in F12 and KLKB1. Previously known (solid) and proposed in this study (dashed) cleavage reactions are represented with arrows. Nominal P values for the associations between protein levels and rs3733402 in the KLKB1 locus and rs1801020 in the F12 locus are shown, respectively, in orange and purple boxes next to the protein of interest.

Figure 4.

Kallikrein cleaves proBNP in vitro. A, A silver stain of recombinant proBNP and kallikrein incubated together for 30, 60, and 90 min with and without a kallikrein-specific inhibitor (PPACK II) and (B) a western blot of an identical experimental setup using an anti-BNP antibody. The silver stain binds all protein present and is a more sensitive procedure than using the anti-BNP antibody for the Western blot. We think that this explains why the amount of proBNP in the +/+/− wells visually seems to be different between the silver stain and Western blot.