Multi-omics combined to investigate potential druggable therapeutic targets for stroke: A systematic Mendelian randomization study and transcriptome verification
Yiming Shi,Le Bao,Yanling Li,Dian Ou,Jiating Li,Xiaodan Liu,Nujiao Deng,Changqing Deng,Xiaoping Huang,Wei Zhang,Huang Ding
DOI: https://doi.org/10.1016/j.jad.2024.08.182
2024-08-28
Abstract:Objective: Stroke is a highly prevalent and disabling disease whose disease mechanisms are not fully understood. The discovery of disease-associated proteins with genetic evidence of pathogenicity provides an opportunity to identify new therapeutic targets. Method: We examined the observed and causal associations of thousands of plasma and inflammatory proteins that were measured using affinity-based proteomic assays. First, we pooled >3000 relevant proteins using a fixed-effects meta-analysis of 2 population-based studies involving 48,383 participants, then investigated the causal effects of stroke and its subtype-associated proteins by forward Mendelian randomization using cis-protein quantitative locus genetic tools identified from genome-wide association studies of these >48,000 individuals. To improve the accuracy of causal estimation, we implemented a systematic Mendelian randomization model that accounts for cascading imbalances between instruments and tested the robustness of causal estimation through multi-method analyses. To further validate the hypothesis that ginsenoside Rg1 monomer acts on the five protein targets screened for drug-targeted regulation, we conducted a comparative analysis of the mRNA (gene) expression levels of a limited number of genes in the brain tissues of different groups of SD rats. The druggability of the candidate proteins was investigated and the mechanism of action and potential targeting side effects were explored by Phenome-wide MR. Results: Six circulating proteins were identified to have a significant genetic association with stroke (PFDR < 0.05). For example, in patients with cardioembolic stroke, higher genetically predicted APRT was associated with a lower risk of cardioembolic stroke (ORivw [95 % CI] = 0.641 [0.517, 0.795]; P = 5.25 × 10-5, ORSMR [95 % CI] = 0.572, [0.397, 0.825], PSMR = 0.003). Mediation analyses suggested that atrial fibrillation, angina pectoris, and heart failure may mediate the association of CD40L, LIFR, and UPA with stroke. Molecular docking revealed promising interactions between the identified proteins and glycosides. Transcriptomic sequencing in animal models indicated that ginsenoside Rg1 may act through APRT, IL15RA, and VSIR pathways, with APRT showing significant variability in mRNA sequencing expression. Phenome-wide MR of the six target proteins showed an overwhelming predominance of PFDR > 0.05, indicating less toxicity. Conclusions: The present study provides genetic evidence to support the potential efficacy of targeting the three druggable protein targets for the treatment of stroke. This is achieved by triangulating population genomic and proteomic data. Furthermore, the study validates the pathway mechanisms by which APRT, IL15RA, and VSIR dock ginsenoside Rg1 in animal models. This will help to prioritize stroke drug development.