Mining the Plasma Proteome for Insights into the Molecular Pathology of Pulmonary Arterial Hypertension

Abstract

Rationale: Pulmonary arterial hypertension (PAH) is characterized by structural remodeling of pulmonary arteries and arterioles. Underlying biological processes are likely reflected in a perturbation of circulating proteins. Objectives: To quantify and analyze the plasma proteome of patients with PAH using inherited genetic variation to inform on underlying molecular drivers. Methods: An aptamer-based assay was used to measure plasma proteins in 357 patients with idiopathic or heritable PAH, 103 healthy volunteers, and 23 relatives of patients with PAH. In discovery and replication subgroups, the plasma proteomes of PAH and healthy individuals were compared, and the relationship to transplantation-free survival in PAH was determined. To examine causal relationships to PAH, protein quantitative trait loci (pQTL) that influenced protein levels in the patient population were used as instruments for Mendelian randomization (MR) analysis. Measurements and Main Results: From 4,152 annotated plasma proteins, levels of 208 differed between patients with PAH and healthy subjects, and 49 predicted long-term survival. MR based on cis-pQTL located in proximity to the encoding gene for proteins that were prognostic and distinguished PAH from health estimated an adverse effect for higher levels of netrin-4 (odds ratio [OR], 1.55; 95% confidence interval [CI], 1.16-2.08) and a protective effect for higher levels of thrombospondin-2 (OR, 0.83; 95% CI, 0.74-0.94) on PAH. Both proteins tracked the development of PAH in previously healthy relatives and changes in thrombospondin-2 associated with pulmonary arterial pressure at disease onset. Conclusions: Integrated analysis of the plasma proteome and genome implicates two secreted matrix-binding proteins, netrin-4 and thrombospondin-2, in the pathobiology of PAH.

Keywords: Mendelian randomization; case-control studies; genome; protein quantitative trait loci.

Figure 1.

Proteomics and genomics analytic workflow leading to the Mendelian randomization (MR) studies. FDR = false discovery rate; GWAS = genome-wide association studies; NET4 = netrin-4; PAH = pulmonary arterial hypertension; pQTL = protein quantitative trait loci; TSP2 = thrombospondin-2.

Figure 2.

Protein association with pulmonary arterial hypertension (PAH) compared with healthy controls. A volcano plot of aptamers representing 208 independent proteins that met false discovery rate (FDR) (q < 0.05; blue dots) in both discovery and replication analysis. Dependent proteins were no longer associated after correcting for renal function or anticoagulation therapy. “Not replicated” proteins only met FDR in the discovery analysis, and those “Not associated” failed to meet FDR in discovery analysis. “Selected” proteins highlight proteins of interest in the pathology of PAH, which also met “independent” criteria.

Figure 3.

Venn diagram shows the overlap between proteins that discriminate patients with pulmonary arterial hypertension (PAH) from healthy control subjects, proteins associated with prognosis in PAH, and protein instruments for Mendelian randomization studies. pQTL = protein quantitative trait loci.

Figure 4.

A forest plot demonstrating shared effect directions of Mendelian randomization analyses on 3 protein quantitative trait loci performed separately in 4 genome-wide association study (GWAS) substudies on pulmonary arterial hypertension (PAH) contributing to the GWAS meta-analysis. An odds ratio above 1 indicates an increasing risk for PAH with an increase in protein level. BHFPAH = British Heart Foundation PAH study; NIHRBR = United Kingdom National Institute for Health Research BioResource Rare Diseases study; PAHB = U.S. National Biological Sample and Data Repository for PAH, also known as the “PAH Biobank”; PHAAR = Pulmonary Hypertension Allele-associated Risk study.

Figure 5.

Genetic risk for pulmonary arterial hypertension (PAH) inferred from levels of netrin-4 (NET4), thrombospondin-2 (TSP2), and endoglin (ENG) in Mendelian randomization analysis. (A) Box plots of NET4 levels in controls and patients with PAH stratified by genotype at rs17288108, (B) TSP2 levels by genotype at rs73043857, and (C) ENG levels by genotype at rs651007. Protein levels were z-scored to mean and standard deviation in controls for ease of comparison. Color scheme indicates genetic risk for PAH inferred from Mendelian randomization. P values provide the strength of association between protein levels and genotypes in patients with PAH. HC references distribution in healthy controls.

Figure 6.

Increased NET4 (Netrin-4) protein expression in human idiopathic pulmonary arterial hypertension (IPAH) lungs. Immunohistochemical NET4 staining in human non-PAH donor (upper panel) and IPAH (lower panel) lung sections, demonstrating increased NET4 protein expression in distal pulmonary vessels of patients with IPAH. n = 3 patients and 3 donors. Scale bars, 20 μm.

Figure 7.

(A and B) Thrombospondin-2 (TSP2) and netrin-4 (NET4) plasma levels in paired samples from relatives (n = 8) who developed pulmonary arterial hypertension (PAH) during follow-up compared with those (n = 15) who remained healthy. (C) Relationship of change in protein level from the diagnostic visit to mean pulmonary artery pressure (mPAP) measured at diagnostic catheterization. P values in A and B based on two-way repeated measures ANOVA (Pgroup) with Šídák's multiple comparisons test (Padj). R-squared and P value in C derived from linear regression models.