作者
Pang Yao,Mohsen Mazidi,Alfred Pozarickij,Andri Iona,Neil Wright,Kuang Lin,Iona Y. Millwood,Hannah Fry,Christiana Kartsonaki,Yiping Chen,Ling Yang,Huaidong Du,Daniel Avery,Dan Schmidt,Dianjianyi Sun,Pei Pei,Jun Lv,Canqing Yu,Michael Hill,Derrick Bennett
摘要
BACKGROUND: Analyses of genomic and proteomics data in prospective biobank studies in diverse populations may discover novel or repurposing drug targets for stroke. METHODS: We extracted individual cis -protein quantitative trait locus for 2923 proteins measured using Olink Explore panel from a genome-wide association study in prospective China Kadoorie Biobank and UK Biobank, both established ≈20 years ago. These cis -protein quantitative trait loci were used in ancestry-specific 2-sample Mendelian randomization analyses of ischemic stroke (IS) in East Asians (n=22 664 cases) and Europeans (n=62 100 cases). We further undertook colocalization analyses to examine the shared causal variants of cis -protein quantitative trait locus with stroke, along with various downstream analyses (eg, phenome-wide association study, drug development lookups) to clarify mechanisms of action and druggability. RESULTS: In Mendelian randomization analyses, the genetically predicted plasma levels of 10 proteins were significantly associated with IS in East Asians (n=2) and Europeans (n=9), with 6 proteins (FGF5 [fibroblast growth factor 5], TMPRSS5 [transmembrane protease serine 5], FURIN, F11 [coagulation factor XI], ALDH2 [aldehyde dehydrogenase 2], and ABO [histo-blood group ABO system transferase]) showing positive and 4 (GRK5 [G protein-coupled receptor kinase 5], KIAA0319 [dyslexia-associated protein KIAA0319], PROCR [endothelial protein C receptor], and MMP12 [macrophage metalloelastase 12]) showing inverse associations, all directionally consistent between East Asians and Europeans. Colocalization analyses provided strong evidence (posterior probabilities for the H4 hypothesis ≥0.7) of shared genetic variants with IS for 9 out of 10 proteins (except ABO). Moreover, 8 proteins were also causally associated, in the expected directions, with systolic blood pressure (positive/inverse: 4/2), low-density lipoprotein cholesterol (1 positive), body mass index (1 inverse), type 2 diabetes (2/1), or atrial fibrillation (3/1). Phenome-wide association study analyses and lookups in knock-out mouse models confirmed their importance for IS or stroke-related traits (eg, hematologic phenotypes). Of these 10 proteins, 1 was not druggable (ABO), 3 had known primary (F11) or potentially repurposed (ALDH2, MMP12) drug targets for stroke, and 6 (PROCR, GRK5, FGF5, FURIN, KIAA0319, and TMPRSS5) had no evidence of any drug targets. CONCLUSIONS: Proteogenomic investigation in diverse ancestry populations identified the causal relevance of 10 proteins for IS, with several being potentially novel or repurposed targets that could be prioritized for further investigation.