Opposing genetic pathways link obesity to depression through distinct circulating proteins and brain phenotypes

Obesity and depression frequently co-occur, and evidence supports a causal relationship. However, obesity is heterogeneous, and its genetic components may exert divergent, even opposing, effects on depression risk. We clustered BMI-associated single-nucleotide polymorphisms (SNPs) based on their Mendelian randomization (MR) estimates on depression risk and validated each cluster using cluster-specific polygenic risk scores (PRS) in the UK Biobank (N = 457,929). Two-step MR was used to assess whether circulating proteins and brain imaging-derived phenotypes (IDPs) mediated the cluster-specific effects. Four SNP clusters associated with elevated BMI showed differential associations with depression risk: C1 (high-risk) and C2 (moderate-risk) were associated with increased risk, C3 exerted a protective effect, and C4 showed no robust association. In the UK Biobank, C1-PRS (OR = 1.67, 95% CI = 1.56–1.79) and C2-PRS (OR = 1.25, 95% CI = 1.16–1.35) were associated with increased depression risk, whereas C3-PRS was associated with decreased risk (OR = 0.78, 95% CI = 0.71–0.88). Two-step MR identified ANGPTL2, MEP1B, and VWC2L as shared circulating mediators of C1 and C2, with PHLDB1 specific to C1, whereas FURIN, HLA-E, HSPB6, and PARP1 mediated the protective C3 effect. rfMRI connectivity (ICA100 edge 73) and right putamen signal intensity mediated the C1 (1.8%, P = 0.045) and C3 (2.1%, P = 0.014) pathways, respectively. Obesity influences depression risk through distinct, even opposing, genetic pathways mediated by specific circulating proteins and neuroimaging phenotypes, highlighting potential targets for subtype-specific intervention.