Profiling of the plasma proteome across different stages of human heart failure

Abstract

Heart failure (HF) is a major public health problem characterized by inability of the heart to maintain sufficient output of blood. The systematic characterization of circulating proteins across different stages of HF may provide pathophysiological insights and identify therapeutic targets. Here we report application of aptamer-based proteomics to identify proteins associated with prospective HF incidence in a population-based cohort, implicating modulation of immunological, complement, coagulation, natriuretic and matrix remodeling pathways up to two decades prior to overt disease onset. We observe further divergence of these proteins from the general population in advanced HF, and regression after heart transplantation. By leveraging coronary sinus samples and transcriptomic tools, we describe likely cardiac and specific cellular origins for several of the proteins, including Nt-proBNP, thrombospondin-2, interleukin-18 receptor, gelsolin, and activated C5. Our findings provide a broad perspective on both cardiac and systemic factors associated with HF development.

Fig. 1. Sampling of study cohorts from different stages of heart failure.

Schematic representation of the natural history of heart failure (HF) progression, from the initial timepoint of disease onset through a gradual decline with increasing episodes of worsening typically necessitating in-hospital care (decompensations) towards terminal pump failure. Cohorts for the current study were obtained from four stages: before disease onset (population-based cohort), during manifest HF, and serially in advanced HF before and after heart transplantation. Proteomic profiles in subjects with incident HF from the population-based cohort were compared to subjects without incident HF. Manifest HF patients were compared to the population-based cohort. Samples from heart transplant recipients were compared to the corresponding samples from the same patient before transplantation.

Fig. 2. Plasma protein abundance across heart failure stages.

Box plots depicting distribution of (a) CRP, (b) CXCL13, (c) Nt-proBNP, (d) PRKACA, (e) Protein C, and (f) TSP2 across cohorts representing different stages of heart failure. In addition to association with incident heart failure, these six proteins changed markedly (>50%) after heart transplantation (HTx). Cohorts include individual human subjects from a population-based cohort who did not develop heart failure during follow-up (“Population”, n = 583), from the population-based cohort who did develop heart failure (“Incident HF”, n = 185), subjects with manifest heart failure (“Manifest HF”, n = 84), and serial measures from subjects with advanced heart failure before and after heart transplantation (“Before Htx” and “After Htx”, n = 30 for both). All except CXCL13 changed in directional consistency with the risk estimate for incident HF towards the general population. The box center line indicates the median, box bounds represent 25th and 75th percentiles, and the whiskers are the most extreme values within 1.5× the interquartile range from the nearer quartile. RFU relative fluorescence units.

Fig. 3. Association of complement system components with development of heart failure.

Diagram of protein interactions in the complement system based on a published review. The three activation pathways (red borders) lead to activation of complement factor 3 which amplifies the signal via feedback loops, and further activates the two common effector pathways: the anaphylatoxins and the terminal complement complex (blue dashed borders). Directionality of significant associations (p < 0.05) with incident heart failure (HF) from Cox proportional hazards regression models adjusting for age and sex (n = 583 population-representative controls, 185 cases) is indicated for each component with color codes, where red indicates higher concentration with incident HF, green lower concentrations, and yellow no significant association. White color indicates that the component was not present on the aptamer-based proteomics platform.

Fig. 4. Association of coagulation system components with development of heart failure.

Diagram of protein interactions in the coagulation system based on a published review. The two activation pathways (red borders) lead to activation of factor X which further activates II (thrombin) and subsequently fibrin and XIII, which form the bloodclot. The bloodclot is digested into d-dimers by the plasmin system, and three additional systems (blue dashed borders) negatively regulate clotting (protein C, AT, TFPI). Directionality of nominally significant associations (p < 0.05) with incident heart failure (HF) from Cox proportional hazards regression models adjusting for age and sex (n = 583 population-representative controls, 185 cases) is indicated for each component with color codes, where red indicates higher concentration with incident HF, green lower concentrations, and yellow no significant association. Nominally significant associations (p < 0.10) are indicated in orange for higher (two components) and mint green for lower (one component) concentration. All associations were examined only after exclusion of subjects on oral anticoagulation therapy (warfarin). White color indicates that the component was not present on the aptamer-based proteomics platform. APC activated protein C, AT antithrombin III, Dd d-dimer, PC protein C, Pl plasmin, Plg plasminogen, PS protein S, Tbm thrombomodulin, TFPI tissue factor pathway inhibitor, TF tissue factor.

Fig. 5. Association of regulatory genetic variants in cis and trans with plasma proteins.

Regional association plots of genetic loci in a cis-associated with thrombospondin-2, and b in trans-associated with C5a at the CFH locus. Plots depict all single nucleotide polymorphisms (SNPs) within ±500 kb of the lead SNP. Circles indicate SNPs with genomic position on the x axis (human genome build 19), p value on the left-hand y axis and recombination rate on the right-hand y axis. Colors indicate pairwise linkage disequilibrium of each SNP with the lead SNP at each locus.