Proteome-wide Mendelian randomization identifies candidate causal proteins for cardiovascular diseases

Abstract

AbstractBackgroundCardiovascular diseases (CVD) remain the leading cause of death worldwide, while a lack of clarity on underlying mechanisms has hindered development of novel therapies. Integration of human genetics and proteomics across different ancestries can provide novel, affordable, and systematic approach for target identification and prioritization.MethodsMendelian randomization approach was applied to unravel causal associations between 2,940 circulating proteins and 21 CVD. Genome-wide summary statistics from the largest genetic mapping of human plasma proteome and meta-analyses on CVD across FinnGen, UK Biobank and Biobank Japan were used. Forward and reverse causation were studied to distinguish respective targets and biomarkers. Genetic instruments for Europeans and East Asians were derived separately and applied to the cardiovascular outcomes in cohorts from corresponding ancestries. We further prioritized drug targets by integrating biological, clinical and population study evidence from cross-database annotations and literature review. Single-cell enrichment analysis and phenome-wide causality scan were performed to further understand target mechanism of action.ResultsWe found 221 novel candidate causal proteins that impacted risk of one or more CVD through forward MR, and 16 biomarkers whose expression levels were affected by CVD through reverse MR (Bonferroni-adjustedP-value <= 0.05). Forward and reverse MR found largely non-overlapping proteins among CVD (only 2 overlapped: LGALS4 and MMP12), suggesting distinct proteomic causes and consequences of CVD. Many of the candidate causal proteins (73.4%) identified are supported by strong literature evidence for a role in immune response and atherosclerotic lesion formation, angiogenesis and vascular remodeling, myogenesis and cardiac progenitor cell differentiation, and energy metabolism. Single cell integration further prioritized ADAM23 for cardiomegaly, PAM for stable angina pectoris and ventricular arrythmia and LPL for peripheral artery disease, whose transcript expression were enriched in cardiomyocytes. Three protein functional groups were highlighted in the phenome-wide scan for their specific enrichment for CVD, including blood coagulation and fibrinolysis, angiogenesis and vascular remodeling, and cell proliferation and myogenesis.ConclusionsOur study identified potential therapeutic targets for CVD and distinguished them from biomarkers due to reverse causation. This study provides human genetics-based evidence of novel candidate genes, a foundational step towards full-scale causal human biology-based drug discovery for CVD.