Integrative Proteogenomics and Single-Cell Transcriptomics Prioritize Candidate Causal Proteins and Therapeutic Targets in Age-Related Macular Degeneration
TL;DR
By combining large-scale plasma proteomic genetics with single-cell mapping and experimental validation, this study identifies a prioritized set of candidate causal proteins linking neuroimmune activation, vascular remodeling, and mitochondrial bioenergetics in AMD, providing candidate entry points for mechanistic and therapeutic exploration.
Abstract
Age-related macular degeneration (AMD) is a major cause of irreversible visual impairment, yet identifying effector proteins and tissue-specific mechanisms underlying genome-wide association study (GWAS) loci remains challenging. This study aimed to systematically prioritize candidate causal circulating proteins and delineate their cellular and transcriptional dynamics in AMD. We integrated plasma protein quantitative trait loci (pQTL) summary statistics from the UK Biobank Pharma Proteomics Project (UKB-PPP; N=53,022) with FinnGen AMD GWAS data using proteome-wide association studies (PWAS), summary-data-based Mendelian randomization (SMR) with the HEIDI test, and Bayesian colocalization analysis. Prioritized candidates were mapped across human and murine retinal single-cell/single-nucleus RNA sequencing atlases. Transcriptional responsiveness was validated in an independent clinical microarray dataset (GSE103060) and in human retinal pigment epithelial cells (ARPE-19) via in vitro inflammatory stimulation and RT-qPCR. Target tractability was assessed using pharmacological databases. Multi-stage genetic screening prioritized five candidate proteins stratified into two confidence tiers: three Tier 1 causal drivers supported by colocalization (PP4 > 0.80)—including risk factors CSF2, IL20RB, and WARS1 (also known as WARS)—alongside two Tier 2 candidates supported by SMR and HEIDI, comprising risk factor PILRA and inversely associated metabolic factor ACADSB. Retinal transcriptomic mapping localized PILRA specifically to microglia, ACADSB to inner retinal neurons, and WARS1 to photoreceptors, RPE, and vascular compartments, while IL20RB and CSF2 exhibited low baseline expression. In independent validation cohorts, IL20RB and WARS1 were significantly up-regulated in choroidal neovascularization (CNV) membrane-derived RPE from patients with AMD (p<0.01). Exposure of ARPE-19 cells to TNF-α markedly induced mRNA levels of IL20RB (P=0.0025) and WARS1 (p<0.0001). Dual normalization against ACTB as a secondary internal reference yielded consistent significant induction. Pathway enrichment highlighted cytokine-driven receptor cascades (JAK-STAT signaling) and mitochondrial substrate catabolism (branched-chain amino acid and fatty acid metabolism). Drug–target profiling identified small molecules and nutraceuticals interacting with ACADSB, CSF2, and WARS1. By combining large-scale plasma proteomic genetics with single-cell mapping and experimental validation, this study identifies a prioritized set of candidate causal proteins linking neuroimmune activation, vascular remodeling, and mitochondrial bioenergetics in AMD, providing candidate entry points for mechanistic and therapeutic exploration.