A systems biology approach uncovers novel disease mechanisms in age-related macular degeneration

**Background:** Age-related macular degeneration (AMD) is a leading cause of blindness affecting 200 million people worldwide. Despite known genetic risk factors, the molecular mechanisms driving AMD onset and progression remain poorly understood, partly due to the lack of comprehensive molecular data from human macular tissues, especially the retinal pigment epithelium (RPE) and choroid. This study aimed to create a systems biology resource integrating transcriptomics, epigenomics, and genetics to uncover novel disease mechanisms.

**Methods:** The authors generated a multi-omic molecular atlas from well-phenotyped human donor eyes. Bulk RNA sequencing (RNA-seq) was performed on macular and peripheral RPE/choroid from 85 unique donors across normal, early AMD (eAMD), intermediate AMD (iAMD), geographic atrophy (GA), and neovascular AMD (NEO) stages. DNA methylation was profiled using Illumina 850K EPIC arrays on macular RPE/choroid from 82 donors. Single-nucleus RNA-seq (sNuc-seq) was performed on 164,399 nuclei from retina, RPE, and choroid of 7 control and 6 advanced AMD donors. Single-nucleus ATAC-seq (snATAC-seq) was performed on 125,822 nuclei from contralateral eyes of 7 control and 5 AMD donors. Differential expression and methylation analyses were conducted using voom/limma and minfi, respectively. Single-cell data were analyzed using Seurat and ArchR. Integration with AMD GWAS loci was performed using SCAVENGE and peak-to-gene correlation. Rare variant burden testing for WNT pathway genes was conducted in 1,707 GA cases and 2,611 controls from clinical trials.

**Key Results:** In bulk RPE/choroid, 408 DEGs were found in normal vs. eAMD, 886 in iAMD, 719 in GA, 696 in NEO, and 1,001 in pooled dry AMD (FDR<5%, fold change>1.5). No DEGs were found in peripheral tissues. HTRA1 expression was significantly higher in iAMD, GA, and NEO macula. DNA methylation analysis identified 23 genome-wide significant DMPs in normal vs. GA, with 19 showing increased methylation in disease. Single-nucleus RNA-seq revealed a distinct AMD Müller glial state (Müller cluster 3) comprising 80% AMD-derived nuclei, separate from normal (basal) and gliotic states. Pseudo-bulk differential expression identified top genes including CRYAB and CLU, validated by in situ hybridization showing striking upregulation in GA lesions. Peak-to-gene analysis linked 22 genes to 7 AMD GWAS loci, including HTRA1 (R=0.34, FDR=3.47E-14) and C6orf223 (R=0.92, FDR=1.34E-202). Two WNT pathway regulators, FRZB and TLE2, were both differentially expressed and differentially methylated in GA. Rare variant burden testing showed TLE2 was associated with GA risk (OR=2.64, adjusted p=0.009).

**Clinical Implications:** This comprehensive molecular atlas provides a critical resource for understanding AMD pathogenesis. The identification of an AMD-specific Müller glial state distinct from gliosis has implications for retinal regeneration strategies. The discovery of WNT signaling dysregulation, particularly FRZB and TLE2, as novel mechanistic players in dry AMD opens new therapeutic avenues. The peak-to-gene analysis strengthens the case for HTRA1 as the causal gene at the ARMS2/HTRA1 locus and identifies C6orf223 as a potential independent contributor at the VEGFA locus. These findings highlight the power of integrative systems biology for elucidating disease mechanisms and prioritizing therapeutic targets in AMD.