Abstract
Alzheimer's Disease (AD) is the leading cause of dementia globally, significantly impacting health and socio-economic sectors. Our study aims to elucidate the molecular basis of AD using an integrated multi-omics approach. We utilized a combination of genomics, transcriptomics, and proteomics data from multiple tissues (blood, cerebrospinal fluid and brain). Summary-data-based Mendelian Randomization (SMR), colocalization analysis and Heterogeneity in Dependent Instruments (HEIDI) analyses were conducted to establish causality between genetic variants and AD risk. Our results identified causal relationships across multiple omics layers, with significant findings for the Angiotensin-converting enzyme (ACE) and CD33 molecule (CD33) genes. For ACE, our analyses across methylation, expression, and protein levels revealed an overall odds ratio (OR) indicating a protective effect against AD. Specifically, increased methylation at cg04199256 and cg21657705 was associated with higher ACE expression. CD33 demonstrated an increased risk of AD (OR = 1.17, 95% CI: 1.09-1.25).Both genes showed strong colocalization signals, with ACE and CD33 having posterior probability values (PP.H4) of 0.99 and 0.95, respectively. The proteins TMEM106B (PP.H4 = 0.96), SIRPA (PP.H4 = 0.92), CTSH (PP.H4 = 0.77), and CLN5 (PP.H4 = 0.92) also showed strong colocalization evidence. At the protein level, genetically predicted higher levels of TMEM106B (OR 1.44, 95% CI 1.24-1.68), SIRPA (OR 1.03, 95% CI 1.02-1.04) and CTSH (OR 1.04, 95% CI 1.03-1.06) was associated with an increased risk of AD; genetically predicted higher level of CLN5 were inversely associated with AD risk (OR 0.69, 95% CI 0.58-0.83). By identifying multiple candidate targets and regulatory axes, our findings offer a valuable resource for prioritizing genes for functional validation and advancing future therapeutic development.